july 2017 vol. 2, issue 1 july 2017† vol. 2, issue 1† $9.95 † sdt01 cover_layout 1 6/16/17...

JULY 2017 • VOL. 2, ISSUE 1 • $9.95 • www.sdtimes.com

SDT01 cover_Layout 1 6/16/17 11:26 AM

http://www.sdtimes.com

SDT01 Full Page Ads_Layout 1 6/14/17 10:39 AM

http://www.textcontrol.com

NEWS 8 News Watch

12 C is for cognitive

15 WWDC: Apple App Store redesigned for the first time

16 IBM’s journey: Building blockchain in the enterprise

19 EA in the cloud drives digital shift

22 Updating the book on AI and games

25 Citizen developers are a necessity

27 State of DevOps report released

31 How data science improves ALM

COLUMNS 7 INDUSTRY WATCH by David Rubinstein

The Times, it is a-changin’

53 GUEST VIEW by Ciaran Dynes

Subscription models fuel innovation

55 ANALYST VIEW by Peter Thorne

Quantifying software quantities

Contents VOLUME 2, ISSUE 1 • JULY 2017

page 28

page 34

page 10

Software Development Times (ISSN 1528-1965) is published 12 times per year by D2 Emerge LLC, 225 Broadhollow Road, Suite 211, Melville, NY 11747. Periodicals postage paid at Hunting ton Station, NY, andadditional offices. SD Times is a registered trademark of D2 Emerge LLC. All contents © 2017 D2 Emerge LLC. All rights reserved. The price of a one-year subscription is US$179 for subscribers in the U.S., $189in Canada, $229 elsewhere. POSTMASTER: Send address changes to SD Times, 225 Broadhollow Road, Suite 211, Melville, NY 11747. SD Times subscriber services may be reached at [email protected].

FEATURESClimbing the IoT data mountain

page 41

Top 10 considerations when planningDocker-based microservices

DevOps: Continuous integration and delivery Getting an end-to-end perspective

SDT01 page 3_Layout 1 6/16/17 11:41 AM

dtSearch.com 1-800-IT-FINDS

The Smart Choice for Text Retrieval®

since 1991

dtSearch’s document filters support popular file types, emails with multilevel attachments, databases, web data

Developers:

faceted search, advanced data

Visit dtSearch.com for

Instantly Search Terabytes of Data

developer products

®

Over 25 search features, with easy multicolor hit-highlighting options

ADVERTISING SALES

CUSTOMER SERVICE

EDITORIAL

www.sdtimes.com

EDITOR-IN-CHIEFDavid Rubinstein

[email protected]

SOCIAL MEDIA AND ONLINE EDITORSChristina Cardoza

[email protected]

Madison [email protected]

SENIOR ART DIRECTORMara Leonardi

[email protected]

CONTRIBUTING WRITERSLisa Morgan, Alexandra Weber Morales, Frank J. Ohlhorst

CONTRIBUTING ANALYSTSRob Enderle, Michael Facemire, Mike Gualtieri, Peter Thorne

[email protected]

ADVERTISING TRAFFICMara Leonardi

[email protected]

LIST SERVICESShauna Koehler

[email protected]

[email protected]

[email protected]

PUBLISHERDavid Lyman978-465-2351

[email protected]

WESTERN U.S., WESTERN CANADA, EASTERN ASIA, AUSTRALIA, INDIAPaula F. Miller925-831-3803

[email protected]

PRESIDENT & CEODavid Lyman

CHIEF OPERATING OFFICERDavid Rubinstein

D2 EMERGE LLC225 Broadhollow Road

Suite 211Melville, NY 11747

www.d2emerge.com


http://www.dtsearch.com


http://www.spreadsheetgear.com


http://www.nsoftware.com

www.sdtimes.com July 2017 SD Times 7

Industry WatchBY DAVID RUBINSTEIN

There’s a saying that goes ‘when one chaptercloses, another one begins.’

This issue of SD Times marks the close of the BZMedia chapter of this publication’s history andopens the chapter on D2 Emerge LLC, a new-agepublishing and marketing company founded by twolong-time members of the SD Times team: the pub-lisher, David Lyman, and the editor-in-chief … me!

We will work hard to maintain the quality of SD Times and build on the solid foundation thathas been built over the past 17 years. Wherever wego, we hear from readers who tell us they look for-ward to each issue, and they say they’re learningabout things they didn’t know they needed to know.And we’re proud of that.

The accolades are certainly nice — and alwayswelcome. Yet, there is nothing more important tous than the stories we tell. Whether putting a spot-light on new trends in the industry and analyzingwhat they mean, profiling the amazing, brilliantpeople behind the innovation in our industry, orhelping software providers tell their unique storiesto the industry, our mission is to inform, enlightenand even entertain.

But, as much as things will stay the same, therewill be some changes. We will look to introduce youto different voices and perspectives from the indus-try, inviting subject matter experts to share theirknowledge and vision of changes in our industry.The exchange of ideas and free flow of informationare the bedrock of our publishing philosophy.

We will somewhat broaden the scope of ourcoverage to include topics that might once havebeen thought of as ancillary to software develop-ment but are now important areas for you to followas silos explode and walls come tumbling down inIT shops around the world.

We will work to improve our already excellentdigital offerings by bettering the user experienceand the way in which we deliver content to you. So,whether you’re reading SD Times on a desktop atwork, or on a tablet at a coffee shop, or even onyour cellphone at the beach, we want you have thesame wonderful experience.

For our advertisers, we will help guide youtoward the best way to reach our readers, whetherthrough whitepapers, webinars, or strategic adplacement across our platforms. And, we will look

to add to an already robust list of services we canprovide to help you tailor your messages in a waythat best suits our readers.

BZ Media was a traditional publishing company,with a print-first attitude (only because thereweren’t any viable digital platforms back in 2000).D2 Emerge offers an opportunity to strike the rightbalance between a digital-first posture and all thatis good about print publishing.

I would be remiss if I didn’t acknowledge BZ Media founders Ted Bahr and Alan Zeichick,who took a cynical, grizzled daily newspaperman andturned him into a cynical, grizzled technology editor.But as I often say, covering this space is never dull.Years ago, I covered sports for a few newspapers, andafter a while, I saw that I had basically seen everyoutcome there was: A walk-off home run, a last-sec-ond touchdown, a five-goal hockey game. The onlything that seemed to change werethe players. Sure, once in a while aonce-in-a-lifetime player comesalong, and we all enjoy his feats.But mostly sports do not change.

Technology, on the other hand,changes at breakneck speed. Aswe worked to acquire SD Times, I had a chance to look back at the first issues we published, and realized just how far we’ve come.Who could have known in 2000, when we were writ-ing about messaging middleware and Enterprise JavaBeans that one day we’d be writing aboutmicroservices architectures and augmented reality?

Back then, we covered companies such as SunMicrosystems, Metrowerks, IONA, Rational Soft-ware, BEA Systems, Allaire Corp, Bluestone Soft-ware and many more that were either acquired orcouldn’t keep up with changes in the industry.

The big news at the JavaOne conference in2000 was extreme clustering of multiple JVMs on asingle server, while elsewhere, the creation of anXML Signature specification looked to unifyauthentification, and Corel Corp. was looking forcash to stay alive after a proposed merger with Bor-laand Corp. (then Inprise) fell apart.

So now, we’re excited to begin the next chapterin the storied (pardon the pun) history of SDTimes, and we’re glad you’re coming along with usas OUR story unfolds. z

The Times, it is a-changin’David Rubinstein is

editor-in-chief of SD Times.

Who could have known in2000... that one day we’d bewriting about microservices

and augmented reality?


SD Times July 2017 www.sdtimes.com8

GitHub releases 2017Open Source SurveyAs open-source software con-tinues to become a critical partof the software industry,GitHub wants to ensure thecommunity understands thepervasive landscape. Theorganization recently releasedan open set of data designed tohelp researchers, data enthusi-asts and open-source memberscomprehend the overall needsof the community.

Some of the major find-ings highlight how valued doc-umentation is to developers,even though it is often over-looked. The open-source dataalso reveals the impact onnegative interactions, howopen source is used by theworld, and who makes up theopen-source community.

Microsoft introducesDraft for Kubernetesapp development Microsoft announced a newopen-source development toolfor cloud-native applicationsthat run on Kubernetes at Core-OS Fest in May. This is the firstDeis announcement since thecompany acquired the Kuber-netes company back in April.

Draft was developed toaddress the complexity andconstraints the developmentcommunity was facing when itcame to working with Kuber-netes.

“Application containershave skyrocketed in populari-ty over the last few years. Inrecent months, Kuberneteshas emerged as a popularsolution for orchestratingthese containers. While manyturn to Kubernetes for itsextensible architecture andvibrant open-source commu-nity, some still view Kuber-

netes as too difficult to use,”Gabe Monroy, lead PM for con-tainers on Microsoft Azure,wrote in a blog post.

Red Hat to acquireCodenvyRed Hat is adding the develop-er tools and containerizedworkspaces provider Codenvyto its portfolio. The companyhas signed a definitive agree-ment to acquire Codenvy.

“Thanks to the increasingpush towards digital transfor-mation and the use of technol-ogy platforms, including apps,as a strategic business advan-tage, the role of the developerhas never been more impor-tant. But, accelerated innova-tion through agile developmentrequires new approaches andtools,” said Craig Muzilla, seniorvice president of applicationplatforms business for Red Hat.

Red Hat plans to make

Codenvy an essential part ofOpenShift.io, its recentlyannounced hosted develop-ment environment for buildinghybrid cloud services. “Whenthe transaction closes, Coden-vy and Red Hat will combineresources to create an agiledevelopment platform forOpenShift-powered applica-tions,” wrote CEO of Codenvy,Tyler Jewel, in a blog post.

Node.js 8 is now live Node.js 8 is now available witha big emphasis on debuggingand developer workflow. TheNode.js 8 release had previ-ously been delayed becausethe the Node.js team wantedto give themselves the optionto ship the Node.js 8.x releaseline with the TurboFan andIgnition pipeline, which wouldbecome the default in V8 5.9.According to Borins, “thiswould allow our next LTS

release line to run on a moremodern compiler + jit pipeline,making backporting easierand giving us a longer supportcontract from the V8 team.”

Red Hat announcesRed Hat EnterpriseLinux 7.4 betaRed Hat is updating its enter-prise Linux platform with newsecurity and compliance fea-tures, automation and animproved admin experience.The company announced thebeta release of Red Hat Enter-prise Linux 7.4.

The solution is designed togive enterprises a foundationto roll out new apps, virtualizeenvironments and createsecure and hybrid clouds.

The latest beta releasefocuses on mission-criticaldeployments and defendingagainst the latest threats withsupport for network bound

NEWS WATCHNEWS WATCH

FORMAC developer Jean Sammet passes awayJean E. Sammet, a computer scientist widely known for developing the programming language For-mula Manipulation Compiler (FORMAC), passed away late last month. Sammet was 89 years old.

Throughout her life, Sammet developed FORMAC, served as the first Association for Comput-ing Machinery (AMC) female president, helped design theCOBOL programming language, and received a number ofawards in the field like the Ada Lovelace Award and the Comput-er Pioneer Award.

Sammet’s career started off in mathematics, and she turnedover to programming in 1955. After she received the 2009 IEEEComputer Society Pioneer Award, she was asked about how shegot involved in the computer field. She said: “In 1955, I was work-ing at Sperry Gyroscope company on Long Island, and I wasdoing mathematical world involving submarines and torpedoes,

and my boss came over to me one day and said ‘Do you know that we have a couple of engineerswho are building digital computers?’ My answer was yes, I didn’t quite known what it meant butyes, and he said ‘Would you like to be our programmer’ and I said what is a programmer?’ andhis answer, and I kid you not, his answer was ‘I don’t know but I know we need one.”

From there, she went on to teach computer programming classes at Adelphi, oversaw a teamof developers for the U.S. Army’s Mobile Digital Computer for Sylvania, worked at IBM, organizedthe first Symposium on Symbolic and Algebraic Manipulation, became a member of the ACMCouncil, and became a fellow of the Computer History Museum (CHM).

“Jean Sammet was a leading figure in the study of computer programming languages. Herwork has been widely recognized as an invaluable record of the origin and development of com-puter languages used since the start of the computing era,” CHM wrote.

SDT01 page 8-9_Layout 1 6/13/17 12:40 PM


disk encryption, enhance-ments to OpenSSL HTTP/2.0,and updated audit capabilities.

In addition, the solutiontargets management andautomation with the include ofRed Hat Enterprise SystemRoles. This inclusion simplifiesthe management and mainte-nance of Red Hat EnterpriseLinux 6 and Red Hat EnterpriseLinux 7-based deployments.

Other features includeimprovements to RAIDTakeover, an update to Net-work Manager, and supportfor new performance co-pilotclient tools.

Google releasesopen-source platformSpinnaker 1.0Google is giving the open-source community anothertool for continuous deliveryand cloud deployments. Googlehas released Spinnaker 1.0, anopen-source multi-cloud con-tinuous delivery platform,which companies can use forfast, safe and repeatabledeployments in production.

Back in November 2015,

Netflix and Google collaborat-ed to bring Spinnaker, arelease management plat-form, to the open-source com-munity. Since that initialrelease, Spinnaker has beenused in several organizationslike Netflix, Waze, Microsoft,Oracle, and Target.

Spinnaker 1.0 is open-source, comes with a rich UIdashboard, and is able to beinstalled on premises, locally,in the cloud, and it can runeither on a virtual machine oron Kubernetes.

N|Solid v2.2 addsNode data integrationIn order to streamline integra-tion of Node.js applicationdata to team workflows,NodeSource released the newversion of N|Solid v2.2, whichallows users to send Nodeapplication data directly toany statsd-compliant system.

N|Solid adds streamlinedintegration with statsd-compli-ant systems, making it easy forteams to send Node applicationdata to existing monitoring orreporting systems. Teams can

also integrate this data withexisting metrics infrastructure.

Mozilla: Political parties agree on sup-porting net neutrality Despite their varying stanceson today’s issues, there is onething that Democrats, Repub-licans and Independents seemto agree on, and that is pro-tecting net neutrality. A recentpoll from Mozilla and researchfirm Ipsos, shows supportacross party lines for net neu-trality, and it reveals that amajority of Americans do nottrust the government to pro-tect Internet access.

Mozilla found that 76% ofAmericans support net neu-trality, with 81% of Democratsand 73% of Republicans infavor of it. Along with thesefindings, Mozilla found mostAmericans do not trust theTrump administration or Con-gress to protect the Internet(78% place no or little trust,according to the poll).

Other findings across thepolitical spectrum include78% believing that equal

access to the Internet is aright, with 88% of Democrats,71% of Independents, and67% of Republicans in agree-ment. When it comes to corpo-rations protecting access toInternet, 54% of respondentsdistrust ISPs.

While the public waits forthe results of the FCC’s deci-sion for net neutrality, Mozillasaid it will continue to workwith Americans to endorse netneutrality. According to theorganization, it has garneredmore than 100,000 signaturesand over 50 hours of voicemailmessages for the FCC. Mozillais also joining Fight for theFuture, Free Press, DemandProgress, and others for a callto al Internet users to defendnet neutrality.

Google, IBM and Lyftannounce Istiofor microservicesGoogle, IBM and Lyft aremerging some of their learnedbest practices around micro -services to create a new open-source project called Istio.Istio was developed to con-nect, manage and securemicroservices. The goal of theproject is to tackle challengesaround resilience, visibility,and security.

Istio is a Layer 7 traffic mon-itoring and control networkdesigned to work with Kuber-netes everywhere, on premiseor in the cloud. Today, develop-ers can manually enable thealpha release of Istio on GoogleContainer Engine.

The service mesh letsdevelopers delegate a lot ofproblems around visibility andsecurity. It also gives develop-ers and teams traffic encryp-tion, and automatic load bal-ancing for HTTP, gRPC, andTCP traffic. z

MapR unveils cloud-scale data fabric MapR-XDAs part of its rebranding effort tobecome a platform company, MapR isextending its Converged Data Plat-form to create a cloud-scale data storeto manage files, objects and contain-ers. The MapR-XD supports any datatype from the edge to the data centerand multiple cloud environments.

“What we are hearing from ourcustomers is that they are looking fora scalable storage platform, or datamanagement platform, that does a lotof things including support for multi-site, data centers, and cloud at theedge in one global namespace,” saidBill Peterson, senior director of indus-try solutions for MapR.



BY MADISON MOORE

Companies today are expandingtheir collaboration efforts beyond

workplace chat groups and creatingsoftware delivery teams through activi-ties like pair programming, or grouphackathons. One longtime softwareengineer and scrum master thinks thatcompanies can go a step further — cre-ating collaborative groups throughimprov.

Wayde Stallmann, who currentlyworks as an agile coach for Asyn-chrony Labs, has convinced compa-nies that the lessons learned from

improv activities can actually go on tocreate great software delivery teams,and more productive and collabora-tive meetings.

Most people are familiar withimprov troupes, or groups of actors thatget together in front of a live audience.Sometimes these groups are completestrangers, and sometimes, like in therecently added Netflix original “Don’tThink Twice,” they are a band ofcomedic friends that let the audienceset the stage for them, working togetherto craft a hilarious story.

When you bring the world of

improv into a meeting, it promotes thesame team player qualities of a troupeperformance, according to Stallmann.He identifies these qualities as collabo-ration, creativity, communication, andtrust.

Every improv session or perform-ance starts with three- or four-minuteactivities. He encourages companiesto use these games during their dailystandups or agile retrospectives. Infact, one of the great books on retro-spectives, titled “Agile Retrospectives:Making Good Teams Great,” byEsther Derby and Diana Larsen, saysthat the secret to retrospectives is toget everyone talking in the first fiveminutes.

“Well, that’s another benefit of thesegames,” said Stallmann. “In the firstfew minutes, everyone talks and hasequal voice. No one dominates thegame which sets the tone that no onedominates the meeting.”

Unlike other improv games, Stall-mann’s strategies are not scenario-based; instead, the activities are morelike warmup games with lessons to belearned. One of the activities Stall-mann starts with is a game calledAlphabet Conversation. The way itworks is one person starts off with astatement or sentence, beginning withthe letter “A.” Then, the next personstarts a sentence with the letter “B,”keeping the conversation going all theway to the letter “Z.”

Start daily stand-up meetings with a 3-minute improv warm-up. These warm-ups prac-

tice the art of Collaboration, Creativity, Communication, and Trust.

BIG OL’ SENTENCE: The first person comes up with a simple and short sentence.Each successive person repeats the sentence, adding another detail. The progressioncould go something like this: “There’s a car.” “There’s a red car.” “There’s a red sportscar.” “There’s a red sports car on fire.” The game goes until the sentence gets too longfor people to repeat, at which time a new sentence must be started.

FREE ASSOCIATION: The first person says a random word. Each subsequent persontries to say a word that has some connection to the previous word. Everyone must listenfor when the topic changes. If someone is stuck they should toss out any random word.

MINISTER’S CAT: Each person uses the next letter of the alphabet and fills in thephrase “The Minister’s cat is a ____ cat”, where blank is replaced with an adjective.

QUESTION: This could be played every day forever. The leader asks a question andeveryone answers in turn. This could be:• What is something in your car that is interesting or unusual?• What’s the oldest thing in your refrigerator?• How would you blow $200? or maybe $200,000?• Describe your perfect week long vacation. Source: Team First Development

A look at some opening acts



Of course, to really show readerswhat the game is like, Stallmann testedit out with SD Times. Here’s a sampleof the activity:

SD Times: After this call, I have to tran-scribe this interview.

Stallman: But, I want to make sure thatI have time for a good lunch.

SD Times: Can’t you go to lunch around12:45 pm?

Stallman: Didn’t really want to do thatbecause I have a masseuse coming inat 12:30 pm.

SD Times: Everything will work out justfine...as soon as I transcribe thisinterview…

Stallman: Fine? You call that fine?! Ohyeah, that is fine, let’s get to work onthat interview right now!

“You and I, as a team, created a storyfrom nothing and neither of us couldhave done it on our own,” said Stall-mann. “We get into the aspect of how ateam can solve a problem that no oneindividually can solve.”

Not everyone approaches these“games” with open minds and arms,according to Stallmann. He has comeinto trainings and had teams that simplysay, “We don’t play games.” Whengreeted like this, he encourages themto try the activities, and by the end of it,they are laughing, they let their guarddown, and they actually learn a skillfrom it, he said.

“When I was AT&T as a Scrum Mas-ter, I had a team where we did a three-minute warmup game to start eachstandup, and we did this for two years

straight,” said Stallmann. “They did itfor a year after I left the team, which Ithink is testament to the fact that it was-n’t because I was asking them to do it.”

Stallmann does emphasize one thing,and that is he doesn't use these activitiesto promote faster software deliveryusing improv. That’s not the connectionhe is making, nor is he saying that doingthese improv activities will ultimatelylead to an “a-ha” moment from softwaredelivery teams. He promotes the use ofimprov activities to create team players,which is exactly what today’s softwaredelivery teams need.

And since creativity is subjective, it’shard to put a metric around thesegames and say that an improv game ledto a breakthrough in software design,said Stallmann. These games are moreabout opening up in the creative senseto help get the team to more creativesolutions and innovations. One thing isfor certain, he said, and that is thatimprov activities do get teams into thecollaborative spirit, it gets themfocused, and it brings teams back “intothe moment to be present.”

Often times in meetings, people’sminds wander, but when these gamesare played, there isn’t a moment thatthey aren’t thinking about the game,Stallmann said.

“You are focused and in themoment, and that being in the momentis where we want our software deliveryteams to be,” said Stallmann. “Themore times we practice, the better weget at it.” z

Agile coach Wayde Stallman, left, leads an improv session of warmup activities during a recent Scrum standup meeting.

Phot

o by

Jas

on T

ice


BY MADISON MOORE

Ask Cookie Monster what is the “Letterof the Day,” and he just might give youthree: IBM.

That’s because IBM and SesameWorkshop, the nonprofit organizationbehind the educational programSesame Street, are work-ing together to prove thatit’s possible to enhanceearly childhood educa-tion experiences with anew cognitive vocabularylearning app.

Built on IBM andSesame’s intelligent playand learning platformand powered by IBMCloud, this ecosystemtaps into IBM Watson’scognitive capabilities andcontent from SesameWorkshop’s early child-hood research, and is thefirst of many future cog-nitive apps and toys thatwill be built on the newplatform.

The Vocabulary Learning App is anintelligent tutoring platform for earlychildhood education. It uses Watson’snatural language processing, patternrecognition, and other cognitive com-puting technologies to refine contentand create personalized experiences foreach child.

“This lends itself to atrue transformation inearly childhood education —enabling deep levels of per-sonalized and adaptive learn-ing globally, through multipleexperiences in both digitaland physical worlds,”

said vice president of development andoffering management at IBM WatsonEducation Chalapathy Neti.

With its cognitive power, the Vocab-ulary Learning App continuously learnswith a child as the child engages it, Netiexplained. Instead of bombarding the

child with words he or she may notunderstand, the app identifies eachindividual student’s ability level. It canidentify words or areas that might needadditional focus, “refining the experi-ence to deliver content that engagesand inspires a child — this ultimately

helps advance the child’svocabulary based on his/heracumen,” said Neti. IBM and Sesame completed

an initial pilot with the Vocabu-lary Learning App at one of

the top urban school dis-tricts, Georgia’s Gwin-

nett County public

schools. The Gwinnett pilot program isthe first time that Sesame Workshopcontent and Watson technology havebeen tested by both students and edu-cators.

According to Neti, IBM collected18,000 feedback points from 120 stu-

dents at Gwinnett. Fromthese data points, IBMfound that the app helpedmany students acquirenew vocabulary words,

like “camouflage” and“arachnid.”

“Not only did theylearn the meaning ofthese new words, theybegan to naturally incor-porate the words in theirconversations throughoutthe classroom,” said Neti.“Furthermore, the pilotshowed that the studentsreally enjoyed learningthrough the videos andwith the Sesame charac-ters; this engagement ledthem to listen more close-

ly and ask more questions.”The teachers involved in the pilot

noted that the app was a beneficialaddition to their classroom, and duringthe pilot, several teachers found thatkindergarteners were able to use chal-lenging words (like arachnid) based onthe progression of words they wereexposed to in a two-week period.

Before rolling out the app to stu-dents and educators around the world,IBM will first expand the pilot programthis fall, and then eventually, the com-pany plans to release similar cognitivelearning tools in the future, like gamesand educational toys, said Neti. z


Students at Georgia’s Gwinnett County public schools play with a new cognitive

vocabulary learning app from IBM and Sesame Workshop.

John

O'B

oyle

/Fea

ture

Pho

to S

ervi

ce fo

r IBM

IBM, Sesame Workshop enhance childhood learning with new vocabulary app

C IS FOR COGNITIVE*

*and cookie!

Sesame Workshop

SDT01 page 12_Layout 1 6/13/17 12:42 PM


http://www.aspose.com

Data Quality Made Easy.Your Data, Your Way.

Start Your Free Trialwww.Melissa.com/sd-times

@

NAME

Our data quality solutions are available

on-premises and in the Cloud – fast, easy

to use, and powerful developer tools,

integrations and plugins for the Microsoft

and Oracle Product Ecosystems.

Melissa provides the full spectrum of data

quality to ensure you have data you can trust.

We profile, standardize, verify, match and

enrich global People Data – name, address,

email, phone, and more.

1-800-MELISSAMelissa Data is now Melissa.See What’s New at www.Melissa.com


http://www.melissa.com


BY CHRISTINA CARDOZA

What would happen if suddenly all ourapplications disappeared and develop-ers stopped developing them? Totalchaos, according to Apple’s WorldwideDeveloper Conference heldlast month.

The conference’s messagewas clear: The world isdepending on applicationdevelopers. To help develop-ers keep the app economyalive within Apple, the com-pany announced new softwareupdates, and a redesign of itsApp Store.

The App Store turns 9 thisyear, and for the first timeever, Apple is changing up thelook and feel of the store. Inthe last nine years, the AppStore has seen 500 million vis-itors weekly, has had more than180 billion apps downloaded, and haspaid out more than $70 billion to devel-opers, with 30% being in the last yearalone.

“People everywhere love apps andour customers are downloading them in

record numbers,” Philip Schiller,Apple’s senior vice president of world-wide marketing, said in a statement.“Seventy billion dollars earned bydevelopers is simply mind-blowing. We

are amazed at all of the great new appsour developers create.”

The new App Store will be designedto give users new ways to discover appsand learn about developers. The storewill feature a new home tab calledToday, and it will feature new dedicatedspaces for games and apps. Other fea-tures coming soon include the ability tophase releases and showcase in-appstore purchases within the App Store.

This new redesign will be a part ofiOS 11. “Today, we are going to take theworld’s best and most advanced mobileoperating system and turn it up to 11,”Tim Cook, CEO of Apple said at theevent last month.

iOS 11 will feature a redesigned appdrawer that makes message apps andstickers more accessible, updates toApple Pay, new Siri capabilities, cameraand photo enhancements, and a newDo Not Disturb while Driving mode.Siri will feature new machine learningand artificial intelligence to betterunderstand and help users. The upcom-

ing operating system will also includeOffload Unused Apps capabilities thatautomatically deletes apps that are notin use, a new control center, and multi-tasking capabilities.

The company is also givingdevelopers a new set ofmachine learning APIs theywill be able to leverage intheir own solutions. Thecompany plans to roll out avision API and a natural lan-guage API. In addition,Apple announced a new aug-mented reality framework,ARKit, that developers canuse to bring computer visionto their solutions and imple-ment and interact withobjects.

Developers can startaccessing iOS 11 today, a pub-lic beta of iOS 11 will be avail-

able later this month, and the release isslated for the fall.

The company’s other operating sys-tem is also getting a number of newupdates with the release of macOS HighSierra. The new macOS will feature bet-ter performance, autoplay blocking,tracking prevention, fullscreen splitviewin mail, h.265 for high-efficiency videocoding, Metal 2, and Metal for VR. Inaddition, the operating system willinclude Apple’s File System (APFS) toprovide a more modern file systemexperience, better performance, andenhanced security. Developers canaccess the latest release now. A publicversion is expected to roll out in a cou-ple of weeks.

Other announcements included anew iPad Pro, Apple Pencil updates,and new drag and drop capabilities, andthe HomePod. HomePod is a smartspeaker, similar to Amazon Echo andGoogle Home, that provides AppleMusic, natural voice interaction, andartificial intelligence technology. z

WWDC: Apple App Store redesigned for the first time

Phot

o: A

pple

Imag

e: A

pple

Tim Cook stressed importance of developers.

The App Store is being redesigned to improve app discovery.



BY CHRISTINA CARDOZA

IBM has embarked on a journey to takeblockchain out of cryptocurrency andbuild blockchain software for the busi-ness. Over the last couple of years, thecompany has developed a permissionedblockchain, IBM Blockchain, to solveone of the biggest problems the bitcoinblockchain didn’t solve — protectingdata privacy in the context of industryand government regulations.

“After looking at bitcoin and theblockchain craze in general, we got thisidea for a blockchain for business. Weput the idea to a test in building out anew style of blockchain that was well-suited for regulated companies that hadto interact and follow rules as well aspass audits if an audit would occur,”said Jerry Cuomo, vice president ofblockchain at IBM.

Through this work, IBM workedwith “network founders” to activateblockchain networks on the company’stechnology. Today, IBM wants to takewhat it has learned along this journeyand help other network founders gofrom concept to active network. Thecompany is launching an IBMBlockchain Founder Accelerator pro-gram to help enterprises and theirdevelopers take blockchain networksinto production faster. Through theaccelerator, IBM will pick eight of thebest ideas, or blockchain networkfounders, to participate in the program.Founders will come from a range ofindustries such as banking, logistics,manufacturing and retail. The programwill provide guidance, support andtechnical expertise for gettingblockchain networks up and running.

“Blockchain is a team sport. Withthe right network of participants collab-orating on the blockchain, the benefitscan be exponential,” said Marie Wieck,general manager for blockchain tech-nology at IBM. “IBM has worked onmore blockchain projects than any oth-

er player in the industry and we under-stand the challenges organizations faceand the resources needed to getblockchain networks right the firsttime. IBM is proactively building solu-tions and entire blockchain ecosystemsacross a broad range of industries andwe are sharing our expertise andresources to help more organizationsquickly set up their networks.”

Cuomo explained the acceleratorprogram consists of different parts:

IBM’s blockchain garage offerings,where it will help founders build outtheir ideas; access to IBM’s assets suchas its secure document store, prove-nance engine, process engine andmember onboarding; and access to thedeveloper team for mentorship, coun-seling, and code review .

According to Cuomo, the interest inthe blockchain is only going to getstronger as time goes on. In a C-Suiteexecutive study, Forward Together, thecompany found one third of 3,000 exec-utives surveyed are or intend to useblockchain in their business. Eight inten of those interested say they aredriven by financial pressures and theneed for new business models.

“With blockchain, everyone is look-ing at the same thing at the same time.This new way of making trusted trans-

actions will spawn new business mod-els, processes, and platforms whereboth producers and consumers can bein a connected ecosystem to create newkinds of value,” said Brigid McDer-mott, vice president for blockchainbusiness development at IBM.

Other findings of the IBM reportincluded 100% of those exploringblockchain expect it to support theirenterprise strategy, 78% of blockchainexplorers are investing to respond to

shifting profits, and 78% of thoseactively using blockchain believe cus-tomers are important to advancing thetechnology.

“I think 2016 was the year ofblockchain experimentation,” said Cuo-mo. “2017 is the year of adoption, andthis year we are seeing networks acti-vated. We will measure the success ofthis year by the number of permis-sioned blockchains we see activated. Aswe get into 2018, we are going to seeboth the growth of those networks andthe interoperation of those networks.”

Through the accelerator program,Cuomo hopes to see the industry startbuilding a best practice for how to starta blockchain business, and a blueprintof how to build a blockchain businessthat will make money for the membersof its ecosystem. z

IBM’s journey: Buildingblockchain in the enterprise



http://www.groupdocs.com


http://www.sparxsystems.com/procloud


BY DAVID RUBINSTEIN

The growth and evolution of businesshas required a growth and evolution inmodeling, which has emerged as a wayto communicate ideas and minimizecomplexity across the enterprise.Modeling is used to get a handle onbusiness processes, database design,code engineering — and even theenterprise itself, through organiza-tional charts, workflows and enter-prise frameworks.

Sparx Systems, whose EnterpriseArchitect modeling platform enablesteams to collaborate on business rules,requirements and more to create UML2.5-based models, last month came outwith Pro Cloud Server, a web-basedplatform that gives all project stake-holders a way to create, review, com-ment on and edit models, diagrams andprocesses from any browser anddevice.

According to Geoffrey Sparks,founder and CEO of Sparx Systems,“The ability to dynamically create, col-laborate and integrate models overmultiple domains and technical plat-forms is a remarkable and highly agile

solution that has the capacity to radi-cally improve the quality, accuracy andeffectiveness of model-based projects.”

Sparx has 580,000 registered usersof Enterprise Architect, and was listedin the Challenger category of the latestGartner Enterprise Architecture Mag-ic Quadrant.

The Pro Cloud Server currentlyboasts three main areas. The first is acomprehensive RESTful API forOSLC (Open Services for Life-cycleCollaboration.) This is the platformthat provides access to the back-endEnterprise Architect repository. Thesecond is WebEA, WebEA builds offthe RESTful API and provides a

mobile interface into the model allow-ing the model to be consumed, com-mented and discussed from any devicewith a web browser. The third is a URLbased approach to connecting to boththese services allowing a secure http(s)connection without the need for data-base specific drivers or configurationsfor each platform.

Having the ability act quickly andcollaboratively reflects a radical changein enterprise architecture, according toa presentation at a Gartner EnterpriseArchitecture Summit conference.Organizations need to address businessdisruptions coming from all angles byembracing digital technologies such ascloud computing, intelligent machinesand more.

“EA success in the digital age isfueled by a spirit of alignment and col-laboration with business and IT stake-holders throughout the enterprise,”Gartner wrote in a summary of thesummit. “EA practitioners must bewilling and able to reimagine tradition-al EA roles and responsibilities anddevelop a new business acumen anddigital skill sets.”

Focus on the business analystAlong with the release of Pro CloudServer, Sparx worked with the Interna-

EA in the cloud drives digital shiftModels that can be shared collaboratively speed creation of business value

INDUSTRY SPOTLIGHT: ENTERPRISE ARCHITECTUREINDUSTRY SPOTLIGHT: ENTERPRISE ARCHITECTURE

Content provided by SD Times and

continued on page 20 >

‘The ability to dynamically create,collaborate and integrate modelsover multiple domains and technicalplatforms is... remarkable.’

—Geoffrey Sparks

With Pro Cloud Server and WebEA, remote users can get a real-time view of project content

within Enterprise Architect, at any time, from any device.

SDT01 page 19,20_new_Layout 1 7/5/17 3:06 PM


tional Institute of Business Analysis tocreate and release a public beta of atoolkit for the Business Analysis Bodyof Knowledge® (BABOK® Guide v3).The model, when implemented withinPro Cloud Server, is designed to givebusiness analysts the ability to collabo-rate on requirements and businessmodels, to provide better business out-comes.

“It is the first time business analysistechniques have been actualized in thisway and we are excited with the out-

come,” said Sparx Systems COO TomO’Reilly. “The solution… has the capac-ity to revolutionize the way the BABOKGuide v3 is applied within the enter-prise.”

The Tools and Techniques forBABOK Guide v3 instructs the profes-sional on implementing business analy-sis standards that offer reliability,repeatability and improved productivity,according to Sparks. Better require-ments capture and management resultin better project outcomes.

Analysts have to reach out to various

departments to assess their needs, thenwrite it out, analyze it and deliver it intechnical language to the developmentteam. Without adequate tooling to han-dle requirements, this task alone canadd months to a software project.

Especially in today’s software devel-opment environment of agile practicesand continuous integration/deliver, theneed to respond quickly to customerfeedback as well as the needs of thebusiness requires that collaborativetools be in place to let stakeholderswork through their needs while provid-ing transparency into the changeprocess.

As Sparks noted in a recent blogpost: “These examples are a reflection ofthe changes in the Requirements Man-agement process, which in the develop-ment model of today, supports iterativerequirements gathering and continuousdelivery of software. It has become anAgile practice approach, being adoptedto address the challenge of digital trans-formation. This collaborative, iterationbased business lifecycle, betweenrequirements and stakeholders, has giv-en rise to DevOps, a strategy for manag-ing continuous change.

Enterprise Architect is unique in itsability to support requirementsthroughout the development lifecycleand to deliver the benefits of the Agilepractice approach. Requirements canbe defined in the model, or importedfrom other tools including Visio.”

An organization using the WebEAview into the Enterprise ArchitectCloud could have a business analyst orproject manager create requirements,use cases, tests and more right in themobile browser. This ability to capturetask assets remotely for use in a modellater on helps make the process moreagile.

More detailed information on Sparx,Enterprise Architect, Pro Cloud Serverand WebEA can be found atwww.sparxsystems.com. z

INDUSTRY SPOTLIGHT: ENTERPRISE ARCHITECTUREINDUSTRY SPOTLIGHT: ENTERPRISE ARCHITECTURE

EA in the cloud helps drive digital transformations< continued from page 19

For a younger generation of developers and business analysts, the term ‘enterprisearchitecture’ might not be totally understood. Yet EA could very well help them – andtheir organizations – better compete in a world that is changing ever more rapidly.

In fact, according to a paper titled “A Common Perspective on Enterprise Architec-ture” produced by the Federation of Enterprise Architecture Professional Organiza-tions, EA was first developed to help companies deal with the shifting, sliding technol-ogy landscape and with diversity in operating systems.

Today, enterprise architecture continues to be implemented, but instead of helping organizations deal with a migration off mainframes onto distributed systems,for example, it is being used to help organizations make a digital transformation.Through the use of models, EA gives adopters repeatable techniques to map out theirfuture – whether that’s a migration to the cloud, or implementing a strategy formobile device access to data – and ensure its success.

According to the paper, “Organizational changes can be dramatic, with large-scalereorganization of people, systems and accountabilities. They can also be gradual andsteady, involving hundreds of small, non-disruptive steps. Regardless of the approachtaken, change is often complex and error-prone. Enterprise Architecture, throughcontinuous evaluation and adaptation of the enterprise, reduces the cost of changeand improves the chances for success.”

–David Rubinstein

What Exactly IS Enterprise Architecture?

SDT01 page 19,20_new_Layout 1 7/5/17 3:06 PM

DON’T MISS A SINGLE ISSUE!

Renew your FREE subscription to SD Times!

Take a moment to visit sdtimes.com. Subscribing today means youwon’t miss in-depth features on the newest technologies affecting enterprise developers — IoT, Artificial Intelligence, Machine Learningand Big Data.

SD Times offers insights into the practices and innovations reshaping software development such as containers, microservices, DevOps and more.

Find the latest news from the software providers, industry consortia, open source projects and research institutions.

Available in two formats — print or e-mail with a link to download the PDF.

Subscribe today to keep up with everything happening in the software development industry!

Sign up for FREE today at www.sdtimes.com.

SDT01 page 21_WirelessDC Ad.qxd 6/16/17 10:28 AM Page 1

http://resources.sdtimes.com/sdtimes-subscription


BY MADISON MOORE

Artificial intelligence research hasseen a lot progress over the years,moving from simply understandingimages and speech to actually detect-ing emotions, driving cars, searchingthe web, and playing games. Becauseof these advancements, two AI expertsdecided to write “Artificial Intelli-gence and Games,” which will theyhope will serve as the first comprehen-sive textbook on application and use ofAI in, and for, games.

Georgios Yannakakis and JulianTogelius, authors of the book, haveboth been teaching and researchinggame artificial intelligence at graduateand undergraduate levels.

Yannakakis is currently an associateprofessor at the Institute of DigitalGames at the University of Malta. Hisresearch interests include everythingfrom AI, affective computing, neuralnetworks, and procedural content gen-eration. Togelius is an associate profes-sor at the Department of ComputerScience and Engineering at the TandonSchool of Engineering at New YorkUniversity. His research interestsinclude artificial intelligence tech-niques for making computer games bet-ter, and he’s researching how games canmake AI smarter.

Both Yannakakis and Togelius feltthat a textbook on game AI was neces-sary to future students and beneficial tothe learning objectives of their pro-grams.

“[Yannakakis] and I have beenworking in this research field since atleast 2005,” said Togelius. “We wereactive at the very beginning of thefield's formation, and did a number ofinfluential research contributions in forexample procedural content generationand player modeling.”

Togelius added that the book is notonly built on their own research, but italso features a lot of research publishedin the IEEE Conference on Computa-

tional Intelligence and Games, theAAAI Conference Artificial Intelli-gence in Interactive Digital Entertain-ment and the IEEE Transactions onComputational Intelligence and AI inGames. They are also looking at workthat's being done in the industry andpresented at conferences like the GameDevelopers Conference.

While there are a couple of booksout there that delve into some of thesetopics around artificial intelligence,Togelius said that these books are older

and tend to focus more on the needs ofthe game industry.

“Our book is fully up to date withacademic research as well,” saidTogelius. “The use of video games asresearch environments in AI researchin academia (and in big companies suchas Google DeepMind and Facebook AIResearch) has exploded in recent years,and our book reflects those develop-ments.”

Togelius also said that unlike otherbooks, he and Yannakakis devote plentyof space to discussing the role of AI incontent generation and player model-ing, not just in game-playing.

“There is an increasing use of video

games to develop and test AI, andthere's also a long-standing use of AImethods in games,” said Togelius.“These fields use very different meth-ods, and don't always understand (oreven know of) each other. While wecome from the academic perspective,we are doing our best to try to bridgethis divide.”

The authors also represent a thirdperspective, and that is “the growinginterest of AI in the game designprocess, to help designers by automat-

ing some of the design or provide auto-mated feedback or suggestions,” saidTogelius.

Right now, the book’s first publicdraft is available for review. Any sug-gestions or feedback will be acceptedby no later than June 20. Togelius andYannakakis’ book will include threemain chapters around playing games,generating content, and modelingplayers.

They will also have an introductorychapter with overviews of the field andsummaries of key algorithms, with afew chapters trying to “stake out thefuture of the research field,” saidTogelius. z

Updating the book on AI and games Professors write comprehensive text reflecting new developments

The “Artificial Intelligence and Games” book includes a set of exercises on ways to use AI in games,

like this exercise from Catlike Coding where you can generate a maze and navigate through it.

Imag

e: J

aspe

r Flic

k, C

atlik

e Co

ding



http://www.xceed.com

LEARN MORE

WWW.REDHAT.COM/INTEGRATE

FOR INTEGRATIONTRUSTED


http://www.redhat.com/integrate


BY LISA MORGAN

Software development team roles arechanging as the pace of business contin-ues to accelerate. Agile development,continuous integration and continuousdelivery continue to become moreimportant. At the same time, there aremore low-code and no-code platformsthat enable less technical “citizen devel-opers” to build, update and enhance lineof business applications.

“We need citizen developers becauselines of business need to keep pace withrapidly changing market conditions andregulatory requirements,” said PrakashAradhya, product management directorat Red Hat. “With all of thosechanges, line of business profes-sionals want more control of theirapplications so they can make thechanges necessary, update thosequickly and get to market faster.”

Toward that end, more ITdepartments are creating infra-structures that help abstract thetechnical complexity of softwaredevelopment so citizen develop-ers can create, maintain and manageline of business applications with drag-and drop simplicity.

Citizen development is growingMore software organizations havemoved away from waterfall develop-ment because their companies can’twait months or years for competitivebusiness applications they need today.Agile and lean development methodshave accelerated software delivery, butthey don’t ensure that all line-of-busi-ness applications are always up-to-date.As a result, business users continue towait for application changes they thinkshould be implemented faster, so moreof them are looking for ways to updateand build applications themselves.

“In an ideal world, IT would set upan app service around some of the exist-

ing infrastructure so it can be con-sumed easily by business applications,”said Aradhya. “The more plumbing ITdevelopers can do, the less coding linesof business have to do.”

Cloud-based services helpCloud-based development models sim-plify service provisioning and theymake it easier for citizen developers toconsume the services their IT depart-ments provide. Citizen developers alsoneed a way to make sense of those serv-ices because they tend not to under-stand software architecture and relatedissues. Low-code and no-code plat-

forms mask all that complexity behindvisual interfaces that citizen developerscan easily understand and use.

Not all low-code and no-code plat-forms integrate equally well into exist-ing business processes, however. If citi-zen developers have to change the waythey work to conform to the limitationsof a particular tool, they’ll either stopusing it or risk losing some of the time-to-market benefits the tool is designedto provide.

“Architectural flexibility is critical intoday’s dynamic environment,” saidAradhya. “Citizen developers should beable to use tools within the context ofexisting business operations.”

The cloud provides both IT andlines of business with other foreseeablebenefits including simple infrastructureprovisioning and elasticity, which are

necessary to speed application changeswhile controlling costs.

Meanwhile, citizen developers arehearing more about the benefits ofmicroservices so they’re starting to askwhether IT and the tools they use sup-port them. Using the cloud, IT can eas-ily make microservices available thatcitizen developers can consume andcombine at will, assuming their plat-form supports them.

Automation speeds processesRobotic Process Automation (RPA) isexpected to enable a lot of businessprocess efficiencies, but many lines of

business are concerned aboutjob displacement. Still, manysoftware development tasks,particularly those that are easi-ly repeatable and reproducibleare already being automated.Additional tasks will be auto-mated in the future that willenable citizen developers toaccomplish more using theirexisting skills.

“Citizen developers aren’t expected tohave a deep understanding of the code,so automation will help simplify businessapplication changes and the creation ofnew applications,” said Aradhya. “Theautomation will range from rote, repeti-tive tasks to more complex and predictivecognitive process automation. Ultimate-ly, there’s an opportunity for lines of busi-ness to identify how they can streamlinetheir operations.”

For now, citizen developers are moreconcerned about timely software deliv-ery which is further enabled by automa-tion and self-service capabilities. As thepace of business continues to accelerate,more lines of business will be demand-ing platforms and tools that enable themto make changes to their own applica-tions quickly and simply.

Learn more at www.redhat.com. z

Citizen developers are a necessity

INDUSTRY SPOTLIGHT: LOW-CODE SOLUTIONSINDUSTRY SPOTLIGHT: LOW-CODE SOLUTIONS


‘Citizen developers aren’t expected to have a deep understanding of the code, so automation will help simplify business applicationchanges and the creation of new applications.’—Prakash Aradhya,

product management director



http://www.leadtools.com

DEVOPS WATCHDEVOPS WATCH


BY DAVID RUBINSTEIN

Automation is an important techniqueused by high-performing IT organiza-tions. Yet medium-performing groupsdo more manual work than low-per-forming groups, according to the 6thannual State of DevOps Report releasedat the DevOps Enterprise Summit inLondon.

The middle performers were doingless automation of processesfor change management, test-ing, deployment and changeapproval, explained AlannaBrown, senior product mar-keting manager at Puppet,which presented the reportalong with DORA (DevOpsResearch and Assessment).“These groups have alreadybegun automation, and areseeing benefits,” she said. “But thatreveals technical debt that they didn’trealize before they started, which is anormal phase of the [DevOps] journey.It’s a J curve; the initial performance ishigh, but then it gets worse before it getsbetter again.”

Another measure of successfulDevOps implementations is leadership.Transformational leaders have a clearunderstanding of the organization’svision, communicate in a way thatinspires and motivates, challenge theirteams through intellectual stimulation,are supportive, thoughtful and caring ofothers, and are generous with praise,according to Nicole Forsgren, CEOand chief scientist at DORA. “It’s hardto measure the impact of leadersbecause they’re not doers or practition-ers, but still they have a big influenceover teams and architecture,” she said.

Looking at the impact of IT perform-ance on overall organizational perform-

ance, the report found — not surprising-ly — that high IT performance (as meas-ured in throughput of code and stabilityof systems) results in organizational suc-cess in faster time to markets, improvedexperiences for the customer and theability to respond quickly to changes inthe market.

The report found that loosely cou-pled architectures and teams result in

higher IT performance. Theadoption of lean product man-agement techniques also fac-tors in. “There is no longer a‘done’ in software,” Puppet’sBrown said. “Teams are work-ing in small batches, makingtheir work visible and usingfeedback to inform design.”

Jez Humble, one of thefounders of DORA, tied in the

ways teams are set up with IT perform-ance and organizational success. “Doesallowing teams to make their own toolchoices and change systems predict anability to do continuous delivery? Canthey do testing on demand, without rely-ing on other teams or services? Doteams have the autonomy to get workdone without fine-grained collaborationwith other teams?” These factors, hesaid, impact whether or not an organiza-tion can ensure their software is alwaysdeployable.

As for system stability and resilience,Humble measures these by how long ittakes to restore the system after an out-age, or what proportion of changes leadto outages or degradation of quality ofservice.

To Humble, the important questionin measuring success of a DevOpsimplementation is, “Can you deploysoftware on demand, during businesshours, at any point of the life cycle?” z

State of DevOps report releasedAutomation, loose coupling of architectures and teams,and leadership are predictors of organizational success

n A newly released study from CA

Technologies reveals if companies

really want to boost their software

delivery performance, they should

combine DevOps with cloud-based

tools and delivery mechanisms. The

study polled more than 900 senior

IT professionals worldwide to deci-

pher how they were achieving

success. According to the study,

81% of respondents saw an overall

improvement when they combined

DevOps methodologies with cloud

tools, compared to 52% who report-

ed an improvement just using

DevOps and 53% when just leverag-

ing cloud.

n New Relic is adding new capabili-

ties to its New Relic Digital Intelli-

gence Platform to give enterprises

better insight into their applications

and infrastructure. The company’s

new Health Map feature provides

all the performance information a

company needs to know in one

unified view. The new feature is

designed to give DevOps teams the

ability to quickly understand if there

is an issue. Other features include

integrated support with AWS strate-

gies and New Relic Infrastructure

enhancements.

n A recently announced testing

solution from Sauce Labs is

designed to help DevOps teams

bring software to market faster. The

Sauce Labs Test Analytics solution

supports DevOps initiatives by pro-

viding near real-time, multi-dimen-

sional test analytics. Key features

include: quick data analysis, early

access to build and test statistics,

the ability to identify errors with

multidimensional filters, ability to

view levels of parallelization, and

REST APIs. In addition, analytics are

available to executives, developers,

test engineers and quality assur-

ance teams.

In other DevOps news…

Nicole Forsgren


SDT01 Full Page Ads_Layout 1 6/16/17 11:55 AM Page 2

http://www.syncfusion.com/SDebook


Businesses are in the midst of adigital transformation. To trans-form, they must become soft-

ware companies, they must turn theirproducts and services online, and theymust provide more intelligent solu-tions. This new and connected worldhas companies turning to the Internetof Things (IoT) to lead them towardsnew business opportunities.

The Internet of Things has beenmaking headlines for years now, butaccording to Bart Schouw, director ofIoT at Software AG, the area is still newand unknown. Businesses are still tryingto navigate and understand how theycan implement this, and how they canthen monetize it.

It is important to distinguish that theInternet and the Internet of Things aretwo very different things. Bryan Hugh-es, CTO of IoT at SpaceTime Insight,an advanced analytics solution provider,explained the Internet is the digital rep-resentation of digital things such as webpages while IoT is the digital represen-tation of physical things.

These physical things or devices pro-vide the ability to better understandand engage with customers, anticipatefailures, and avoid costly downtimes;but to do that businesses need to beable to collect, analyze, store and com-prehend data. Data is the key to makingvaluable insights. The problem in theIoT world is you now have the ability toapply inexpensive sensors onto solu-tions that enable you to collect all kindsof different information. Suddenly, youend up with a mountain of data you caneasily get buried in.

“We collect all this data, and as aresult the data is growing like crazy,”said Svetlana Sicular, research vicepresident for data and analytics at Gart-ner. “There is a big challenge to under-stand which data is useful, and how tomake sense of it. So far, there aren’t too

Climbing the IoT data mountainBY CHRISTINA CARDOZA

SDT01 page 28,29,32,33_Layout 1 6/14/17 3:00 PM Page 28


many companies that are successfulanalyzing this data.”

Treading through deep data waters The biggest change that is enabling theIoT age is changing how data isprocessed and stored, according to JackNorris, SVP of data and applications atMapR Technologies. Norris explains thefirst phase of IoT was focused on deploy-ing devices and getting access to data.Today, we are in the second phase of IoTwhere we are focused on the data itself:How do we collect it appropriately, howdo we analyze it, and how do we act on it.

“You need an approach that can han-dle high scale, high speed and reliabilityall at the same time because it is reallyabout understanding the context of thedata as fast as possible and being able toact in real time,” he said.

The number one best practice togain valuable insight from IoT solutionsis to utilize data analytics, according toGartner’s Sicular. “It is not only aboutinstalling sensors and creating alerts,but it’s about understanding the long-term value of analytics,” she said. “Ana-lytics is what brings the value of anexisting IoT project to the next level.”

Sicular explains there are four typesof ways to approach analytics: Througha platform with analytical capabilities; ageneral purpose analytics tool; do-it-yourself, custom-developed solutions;and a packaged application that solves aparticular use case.

But before you deal with analytics,you need to handle the data itself, andthat includes the speed of data, theamount of data, and new users of thedata. According to Sicular, the firstquestion you need to answer is “Howare you going to store the data?”

Software AG’s Schouw explainedmore IoT platforms are actually livingin the cloud today because the cloudprovides the ability to scale up and scaleout as well as the space necessary tostore information.

Suraj Kumar, vice president and gen-eral manager of PaaS at Axway, added“The reason companies turn to a publicor private cloud is because IoT devicesand the amount of data need a placethat provides high levels of scalability

and elasticity. A cloud-oriented solutionenables both from a compute perspec-tive and a data storage perspective.”

When deciding on a cloud provider,companies need to understand theirbusiness requirements. According toKumar, there are basic security require-ments and compliance requirementsthat need to be considered. Forinstance, healthcare providers need toensure cloud providers have HIPAAcompliance, financial companies needto have compliance policies as well, andgovernment entities need to have fed-eral compliance. “Businesses need toask who has access to the data and datacenter, and what kind of security andcontrols do they have in place,” he said.

In addition to understanding the

business requirements from a cloud per-spective, knowing the business out-comes will help businesses handle all thedata as well as generate insight. DonnaPrlich, chief product officer at Pentaho,a Hitachi Group Company, explainedwhen there are so many different vari-eties of data and information coming infrom different data sources, it can beoverwhelming to know what to look at.When you focus on the business usecase, you are not trying to take every sin-gle data source in. Instead, you are look-ing at the places the data applies to thebusiness outcome, according to Prlich.

“Focusing on business outcomes andwhat you are trying to accomplish,starting small, and growing is going to


Advanced analytics database provider Kinetica wants to tackle the Internet of Things

and its data analytics tsunami with a new graphical processing unit (GPU) approach.

According to the company, the IoT is expected to reach $470 billion in revenue by

2020 from 30 billion or more IoT devices. In order to access and do more things at

the edge, enterprises need to turn their data into actionable insights. Kinetica started out in 2009 in the United States Army Intelligence and Security

Command space. Its mission was to help the US Army and NSA track and capture ter-rorists as well as discover other national security threats. To do that, they needed tocorrelate a number of different data feeds together and come up with a high proba-bility of where a terrorist target would be in real time. To meet their needs and pro-vide fast data with the ability to scale, Kinetica created a new database around paral-lelization utilizing the GPU.

Today, Kinetica is trying to bring the lessons learned from the US Army to theenterprise. “We started as a geospatial and temporal computational engine for anydata that dealt with space and time. Through the years we became a full-fledged,highly available database. Many of the challenges we saw in the military, we are see-ing now in the commercial space as this IoT phenomenon is becoming more and moreprevalent,” said Amit Vij, CEO and founder of Kinetica. By taking traditional databaseoperations and accelerating them with GPUs, Vij explained, the company has beenable to provide 100 times performance on a tenth of the hardware.

According to Vij, a GPU approach provides more than 4,000 coresper device to enable fast, real-time and predictive analytics while tra-ditional CPU-based solutions provide about 16 to 32 cores per device.

Kinetica can be used for connected cars, fleet management, infra-structure, smart grid, customer experience, and just-in-time inventory.The company has worked with the United States Postal Service (USPS) tohelp track carrier movements in real time and become more proficientand accurate in delivering on time. The company was able to do thisbased on personnel, weather and traffic data.

Other features include geospatial data, time-series data, sensordata, structured data, machine learning, deep learning, location-based analytics, geospatial visualization, third-party integration andreal-time visualization.

Using graphic processing units for IoT data and analytics



http://www.saas.hpe.com/software/predictive


BY ALEXANDRA WEBER MORALES

If you’re an agile team, you may still beplanning, developing, testing and deploy-ing by instinct. But what if you bring datascience into the picture? Enter HPEPredictive Analytics, which can surfaceeverything from accurate planning esti-mates in agile projects to efficiencies indefect detection for continuous testing.

SD Times spoke with Collin Chau, a10-year HPE veteran based in Sunny-vale, about how HPE is applying data sci-ence to historical ALM project data.Chau, senior marketingmanager for HPE ALMOctane and PredictiveAnalytics, explained howmachine learning, anom-aly detection, clusteranalysis and other tech-niques improve the foursstages of the lifecycle.

SD Times: Where do you get data to

improve ALM?

Chau: There’s metadata that sits withinthe ALM platform that’s untapped. Ifyou look closely, a lot of this data can beused to accelerate quality applicationdevelopment lifecycle. The customerswe spoke to want to get more out of thatdata, to use it to help guide applicationproject teams optimize resources andreduce risk when managing the applica-tion development lifecycle.

We have an experienced team ofdata scientists today developing algo-rithms for one thing alone: qualityapplication delivery. Data science withthe absence of domain knowledge isuseless information. We have data sci-entists sitting in application develop-ment lifecycle teams to cross-pollinateour tools with ALM-specific data sci-ence, offering users prescriptive guid-ance that is pertinent.

These advanced analytics are multi-variate in nature, and borrow technolo-gies specific to machine learning thatcontinuously learn from past data —because only with a constant learningcycle that feeds on updated data that youimprove and offer better recommenda-tions and predictions.Where do these analytics show up?

Predictive analytics is offered and sold asa “plug-in”, to be shown through theALM Octane dashboard, which we arepositioning as a data hub that will feed

into other ALM tools that are on themarket to offer a single source of truth. Do you have real-world examples?

We have several customers who are par-ticipating in the technology review. Pre-dictive Analytics for ALM will go publicbeta next month, and we’ll have generalavailability shortly. You describe four stages of predictive

analytics in ALM. What’s the first one?

The first is predictive planning. Mostagile projects have no proper planning;teams start out in the dark running as fastas they can. Development time framescan get over-extended or mis-resourced.In predictive planning, the tool learnsfrom past historical data and providesteams the recommendations in terms ofuser requirements, story points, and fea-ture size estimates, etc.Next is predictive development?

For coders, the number-one job is tobuild quality code fast. The tool is intel-ligent enough to predict code that willbreak the build even prior to code

check-in, to proactively analyze sourcecode for defects or complexity. It canalso recommend code to supplementthe build — I’m pretty excited that ithas the ability to continuously learnfrom different data points and classify itinto information that developers canactually use to avoid rework.Stage three is predictive testing —

does that prevent QA from being

squeezed on both ends by DevOps?

A: Yes. It’s about how you acceleratenot just testing, but continuous testing.In this world of continuous delivery,predictive testing gets you to the nextlevel. It helps identify root cause analy-sis in test failures. By so doing, it actu-ally goes a step further by recommend-ing a subset of tests to be done based oncode changes checked-in.

Predictive analytics helps you zoomin. It says you don’t have to run a suite of20,000 automated tests, when 100 specif-ic ones are sufficient enough to cover thelatest code commits.The final stage is predictive operations?

We are taking real-world productiondata and leveraging that to tell the cus-tomers where there are test inefficien-cies. The wow about this is that,because we are taking actual produc-tion data, we are infusing applicationdevelopment decisions with data fromreal-world conditions. It’s no longer lab-based or static, as applications are con-sistently refined to meet needs in actualoperating environments.How can teams try this on for size?

To sign up for the public beta, go tosaas.hpe.com/software/predictive. Learnto optimize your resource investmentsand reduce risk for agile app releaseswithin DevOps practices. Discover howpredictive analytics multiplies the powerof ALM Octane’s data hub as a singlesource of truth. z

How data science improves ALMTeams can tap into data to optimize the four stages of agile application delivery

INDUSTRY SPOTLIGHT: PREDICTIVE ANALYTICSINDUSTRY SPOTLIGHT: PREDICTIVE ANALYTICS


‘The tool is intelligent enoughto predict code that willbreak the build even prior to code check-in.’

—Collin Chau

SDT01 page 31_Layout 1 6/14/17 2:02 PM Page 31


be super important to be successful,”she said.

Rather than trying to provide everysingle piece of information, Axway’sKumar suggests looking at what thecustomer wants and tailor a solution toa particular customer experience.

4 KEYS TO DATA ANALYTICSAn IoT/data analytic approach shouldconsist of four things, according toSpaceTime Insight’s Hughes: • The collection of data • Edge computing • Processing in real-time • Security For collection, it is about building asystem that can withstand the realworld, meaning building systems thatare designed for failure. Edge comput-ing allows you to go from the end pointto the cloud. Then you need to be ableto process everything in real timesomehow, and reduce as many attackvectors as possible, Hughes explained.

“The challenge is that as we movetowards the future, more and morethings will be operating in very remotelocations, or traveling through the air oracross the country on the roads and rails.In most cases, connecting through cellu-lar networks. In these cases, the amountof data generated cannot be transmittedfeasibly to the cloud for processing.Instead, data collection and analyticsneeds to move to the edge, improvinglatency, reliability, and cost,” Hughessaid.

In the end, to provide a successfulsolution, Software AG’s Schouw says topmanagement needs to have close tieswith operations because IoT has a hugeimpact on the business. “If top manage-ment hasn’t bought into it and doesn’tunderstand why it is going to change,operations will never be able to makethose painful decisions to reorganize andrealign the organization along the newbusiness models because organizationsare built to resist change,” he said.

Applying AI and machine learningEven taking into account all of the bestpractices and putting the necessary toolsand platforms in place, it still is almostimpossible to sift through all the data,

especially in real time. Humans aren’talways capable of understanding theright questions to ask, Hughes explained.

“The growth of analytics and busi-ness intelligence has always beenaround knowing the question to ask,and then being able to ask the questionto get the answer,” he said. “In themountains of data, it is not about know-ing the question to ask. It is about dis-covering patterns in the data.”

To discover the patterns, businessesneed to leverage machine learning.According to Schouw, machine learninglearns from the data, discovers patternsthat might not be clear to the humaneye, and deciphers whether or not theuser should act on the data or ignore it.Additionally, it helps make data analyt-ics more of an automated process.

“Even if you have predictive analyt-ics in place to spot a pattern and alertan operator to it, there might be somany alerts and data going on that ahuman doesn't want to be confrontedwith that continuously. Being able tohave artificial intelligence or machinelearning take automated action on it issomething you want to do. That is a bigefficiency gain,” Schouw said.

According to Schouw, artificial intelli-gence is becoming the new UI. Schouwcited Amazon Alexa. A user might beable to tell Alexa to lock the door, butAlexa has to figure out which door tolock, the front door, the back door, thebathroom door? And if you want to lockthe front door and someone is still inside,Alexa can ask if you still want to lock it.“In those complex environments, thequestion is how do you want to interact,and AI will be the way humans want tointeract with it,” he said.

While machine learning and artifi-cial intelligence are providing manyproductivity benefits to IoT organiza-tions, Axway’s Kumar believes this areais still new and expects there will bemany improvements in the future tohave the technology automate certaindecision-making and provide evendeeper insights. “Machine learning andartificial intelligence can drive furtherimprovement in getting insight, helpingwith decisions and automating certaindecisions when it comes to IoT analy-

sis,” Kumar said. Software AG’s Schouw notes

machine learning is not a magic bullet.You can’t just apply machine learningout of the box, connect to things, andexpect it to tell you when things willfail. You need to have an understand-ing, and you need to invest in data sci-entists that can help you build up thatknowledge so you can start applyingthings like predictive maintenance tomachine learning, he explained.

Having tools support your strategy Once you find a place for the data tolive and come up with a strategy, youneed a solution to execute on that strat-egy and apply techniques like machinelearning and artificial intelligence.There is no right answer or single toolto help you magically handle the Inter-net of Things landscape, but there aresome features you can look for in a toolto help make life easier.

When looking for a tool, the firstthing to consider is your data pipeline,according to Pentaho’s Prlich. How areyou going to manage the pipeline andhow are you going to address the differ-ent types of users in your organization.For instance, you might have data sci-entists, ETL engineers, analysts, anddevelopers all working with the data insome shape or form, Prlich explained.

“You want to ask ‘Is this somethingthat can help me solve the end-to-endproblem all the way from data engi-neering through data preparation todata analytics, or is it a siloed event orset of tools?’” said Prlich.

In addition, Prlich explained the solu-tion should be open where team mem-bers can bring in a set of tools or plat-forms that can coexist with one another.“You want to think about what is comingin the future, the tools you are choosing,and if things shift and change, are youprepared to manage that,” she said. Thishelps companies “future proof” them-selves for what is coming next.

MapR’s Norris suggests having a dis-tributed data fabric that can extend tothe edge and intelligently process data.The IoT landscape requires businessesto collect data, aggregate, and learnacross a whole population of devices to

< continued from page 29


understand events and situations. Atthe same time, businesses need toinject intelligence to the edge so theycan react to those events very quickly.According to Norris, enterprises needto be able to converge the differentdata cycles, harness data flows and pro-vide agility. Having a common data fab-ric can help handle all of the data in thesame way, control access to the data,and apply intelligence in a high per-formance and fast way.

Security and privacy aspects Security continues to be a huge chal-lenge with the Internet of Things. Asdevices become more widely used andspread out, a bigger surface area ofattack is created. According to Space-Time Insight’s Hughes, machine learn-ing comes into play here as wellbecause businesses need to be able toperform intrusion detection.

“You can’t fully secure anything. Youneed to be able to understand as quick-ly as possible when there has been abreach. Machine learning comes intoplay for that and can do anomaly detec-

tion to determine whether or not a sys-tem has been breached, and thenrespond to it quickly,” he said.

In addition, security has to be gran-ular, according to MapR’s Norris. Anytime data is moving, it has to beencrypted so it is not easily accessed.There also has to be some intelligenceto how data flows, where the datamoves, and where it is processed.

Axway’s Kumar believes API man-agement plays a big role in IoT and dataanalytics because most devices leverageAPIs in some way or another. An APImanagement solution can help ensurethe data being passed is securelyopened up and transmitted. Howeveran API management solution will onlytake you so far when it comes to securi-ty. In addition, businesses need toensure the policies implemented forAPI management are solid, providegovernance, and enforce a set of corpo-rate security policies that don’t enabledata to be accessed by people who don’tneed to access it, Kumar explained.

“API management combined withbest practices, policy management, and

governance help essentially both secur-ing and putting the data into places thatit needs to for further analysis or stor-age,” he said.

As far as the privacy aspect of all ofthis, there are two pieces of it, accordingto Kumar. There is the user aspect andthe company aspect. From the user per-spective, we typically just go throughand click okay into disclosures when wesign up for a service and connect ourdevices online. Users trust the businessto protect them, or look out for theirbest interest. “People are either open ordon’t have the full knowledge of privacy,so on the business side there are stricterrules they need to follow,” said Kumar.

While businesses typically share datato third parties for further analysis,there are strict data privacy laws on howthe data is stored and who has access toit that businesses need to adhere to inmost cases, Kumar explained.

“People are realizing it is not justabout the devices, and not about thefast proliferation,” said MapR’s Norris.“It is about being able to deploy andleverage IoT effectively.” z


When you think of the Internet of Things, typically smartwatch-es, connected home devices, and smart cars come to mind. Butthe Internet of Things extends to all different kinds of indus-tries. For instance, Caterpillar Marine, a subsidiary of the con-struction and mining equipment provider Caterpillar, recentlyturned to the Internet of Things to gain real-time insight intotheir fleets and ships, and provide better customer experience.

Through the company’s data analytics service developed byESRG Technologies, Caterpillar is able to collect informationfrom sensors on their ships to manage their fleets. The type ofinformation collected predicts machinery failure, allowingCaterpillar to schedule necessary maintenance. This type ofservice can provide massive savings and have a huge impact ona business. “We identify trends toward failure before theybecome alerts,” Jim Stascavage, marine asset intelligencetechnology manager at Caterpillar Marine, said in a case study.“The deviation won’t trigger an alarm, but it should, becausethe trend is starting to go in the wrong direction.”

Caterpillar Marine wanted to go further with their analyticalcapabilities, and uncover trends that could potentially providethem with biggest cost savings or payoffs. Caterpillar choose adata integration and business analytics solution from Pentaho

to help it combine its sensor data with operational data andfind meaningful patterns in the equipment and solutions. Thedata Pentaho was able to collect included things like tempera-ture, pressure, geographical coordinates and geometric angles.“We’re mashing all this data together and trying to figure outwhat it means for the performance of the ship,” Stascavagesaid in the case study. “It’s not simple for even one ship, so youcan imagine how complex it is across an enterprise. There areliterally trillions of data points that need to be evaluated everyyear.”

According to Caterpillar, this new IoT analytical approachwas able to provide better insight into equipment performance,strengthen customer relationships with ROI, and even providesavings in fuel efficiency, unscheduled downtime and environ-mental compliance.

“We see this convergence of the machine-generated databeing able to cut into organizations. Applying the other datasources for context is really what is driving these great busi-ness outcomes. That is what we see in the early IoT market. Itis moving quickly and there is a lot of opportunity to takeadvantage of,” said Pentaho’s Prlich.

How a manufacturing company became an IoT company


Replacing monolithic apps — orbuilding greenfield ones — withmicroservices is a growing con-sideration for development teams

that want to increase their agility, iter-ate faster and move at the speed of themarket. Providing greater autonomyamongst different teams, allowing themto work in parallel accomplishing morein less time, microservices offer codethat is less brittle, making it easier to

change, test and update.Docker containers are a natural fit

for microservices as they inherently fea-tures autonomy, automation, and porta-bility. Specifically, Docker is known forits ability to encapsulate a particularapplication component and all itsdependencies thus enabling teams towork independently without requiringunderlying infrastructure or the under-lying substrate to support every singleone of the components they are using.

In addition, Docker makes it reallyeasy to create lightweight, isolated con-tainers that can work with each otherwhile being very portable. Because theapplication is decoupled from the

underlying substrate, it is very portableand easy to use. Last, it is very easy tocreate a new set of containers; Dockerorchestration solutions such as DockerSwarm, Kubernetes, or AWS ECS makeit easy to spin up new services that arecomposed of multiple containers — allin a fully automated way. Thus Dockerbecomes a natural fit for microserviceswhen creating a microservices substrateon which Docker containers can run.

All that said, there are severalprocess and technology design points toconsider when architecting a Docker-based microservices solution. Doing sowill help avoid costly rework and otherheadaches down the road.


Aater Suleman, Ph.D. is CEO & co-founder at Flux7,an Austin-based IT consultingcompany recognized by AWS

for its expertise in DevOps.

TOP 10ConsiderationsWhen PlanningDocker-BasedMicroservices

BY AATER SULEMAN

SDT01 page 34-36,38,39_Layout 1 6/15/17 4:38 PM Page 34

Process Considerations1.How will an existing microservicebe updated?The fundamental reason developersuse microservices is to speed develop-ment, which increases the number ofupdates they have to perform to amicroservice. To leverage microservicesfully, it is critical that this process beoptimized.

However, there are several compo-nents that make up this process andthere are decisions that come with eachstep in the process. Let us explain withthe help of three examples. First, thereis the question of whether to set up con-tinuous deployment or set up a dash-

board where a person presses a buttonto deploy a new version. The tradeoff ishigher agility with continuous deploy-ment versus tighter governance withmanually triggered deployment.Automation can allow implementationof security with agility and allow bothbenefits to co-exist. Developers need todecide their workflows and whatautomation they require, and where.

Second, it is important for business-es to consider where the actual contain-er will be built. Will it be built locally,pushed and travel through thepipeline? Or will actual code first beconverted into artifacts, and then to aDocker image that travels all the way to

production? If you go with a solutionwhere the container is built in thepipeline, it is important to considerwhere it will be built and what tools willbe used around it.

Third, the actual deployment strate-gy must also be thought through.Specifically, you can update a microser-vices architecture through a blue-greendeployment setup where a new set ofcontainers are spun up and then the oldones are taken down. Or, you can optfor a rolling update as you go throughthe multiple service containers, creat-ing one new container and putting it inservice while you take out one of the





old ones. These decisions are multi-faceted and require consideration ofseveral factors including current flows,the skill levels of operators, and anytechnology inclinations.2. How will developers start abrand new service?Starting a new service is a fundamentalrequirement of microservices. As aresult, the process for starting a brandnew service should be made as easy aspossible. In this vein, an importantquestion to ask is, ‘how will you enabledevelopers to start a new service in aself-service fashion without compro-mising security and governance’? Will itrequire going through an approvalprocess such as filing an IT request? Or,will it be a fully automated process?

While I recommenderring on the side ofusing as much automationas possible, this is definitely aprocess point development teamswill want to think through in advanceto ensure you correctly balance theneed for security, governance andself-service.3. How will services geta URL assigned?This question reallygoes hand-in-hand withstarting a brand newservice. A new URL or subcontext(e.g., myurl.com/myservice) needs tobe assigned to a new service each timeit is created and a process for assigningthem should ideally be automated.Options can include a self-service por-tal for assigning URLs manually or aprocess whereby the URL is automati-cally assigned and pulled from thename of the Docker container and anytags that are applied to the Dockercontainer.

Again, just as with starting a newservice, I recommend erring on theside of using as much automation aspossible — and therefore spend someample time thinking through thisimportant design point well in advance.4. How will container failure bedetected and dealt with?A key requirement of modern infra-structure is that it doesn’t require

‘babysitting’; it can self-heal and self-recover if it goes down. As a result, it isparamount to have a process to detectfailure and a plan for how it will be han-dled when it does occur. For example, itis important to have a pre-definedprocess to detect a container applica-tion that is no longer running, whetherthrough a networking check or log pars-ing. Additionally, there should be adefined process for replacing the con-tainer with a new one as a possible solu-tion. While there are many approachesto this process, the design point is tomake sure that the requirements aremet, ideally via automation.5. How will the code for eachmicroservice be structured?We want a fully automated process forbuilding and deploying new services.

Yet, if the number ofservices is going to be

large, it can quickly becomecumbersome to manage. As a

result, multiple versions of the

process — one for eachservice — should be created. In thesecases, it is imperative that each process iskept homogeneous.

A very important decision in this ishow each microservice is to be struc-tured. For example, the Dockerfileshould always appear in the exact sameplace and whatever is specific to theservice should be contained with theDockerfile. In this way, the process canbe made microservice agnostic. Similar-ly, other files such as a Docker composefile or a task definition for AWS ECSshould consistently be put in the sameplace — across all services — so thatprocesses can run consistently in ahomogeneous fashion.

Technology Considerations6. What tool will be used to sched-ule containers on compute nodes? Schedulers are important tools as they

allocate resources needed to execute ajob, assign work to resources andorchestrators ensure that the resourcesnecessary to perform the work areavailable when needed. There are manytool choices for container orchestration.Those typically considered are: ECS forcustomers in AWS, and Docker Swarmor Kubernetes for those who would likea vendor-agnostic solution. There areseveral angles for organizations toweigh in making this decision includingportability, compatibility, ease of setup,ease of maintenance, the ability to plug-and-play, and having a holistic solution. 7. What tool will be used to loadbalance requests between the con-tainers of the same service?High availability and the ability to havemultiple container services in the envi-ronment make it critical to support morethan one container per microservice.For services that are non-clustered, forexample web-based microservices devel-oped in house, there is need for an exter-

nal load balancer to bal-ance incoming trafficbetween different con-tainers on the sameserver. For load bal-ancers within the sameservice, there are severaloptions — from taking

advantage of AWS ELB in Amazon toopen source tools that can act as loadbalancers such as NGINX or HA Proxy.

This is an important technology deci-sion that should be thoroughly evaluat-ed. Some salient design points to con-sider in your evaluation: requirementsfor session stickiness; the number ofservices you plan to have; the number ofcontainers you have per service; and anyWeb load balancing algorithms youwould like to have. 8. What tool will be used to routetraffic to the correct service?This design point goes hand-in-handwith load balancing as it directlyaddresses application load balancing.As pointed out earlier, individual URLsor sub contexts are assigned per service.When traffic hits the microservicescluster, another task is to ensure thatthe traffic coming in is routed to the



A very important decision in this is howeach microservice is to be structured.



http://www.roguewave.com/microservices

BY MOSHE KRANC

Microservice architectures are all therage these days, and with good reason. Ina nutshell, microservices is a softwarearchitecture pattern which decomposesmonolithic applications into smaller sin-gle-purpose services that are built andmanaged independently. The benefits ofa microservices architecture are:n Quality: separation of concerns min-imizes the impact of one service’s bugson another service.n Agility: upgrading an existing serviceor adding a new one can be done on agranular level, without impacting otherservices.n Scalability: Each service can be pro-visioned individually so it has the hard-ware and software resources it needs.n Reuse: Services can be shuffled andcombined in new ways to provide newfunctionality.

As CTO of Ness Digital Engineer-ing, I frequently meet customers whowant to implement a microservicesarchitecture, so they can enjoy all thesebenefits. But, I’ve learned that amicroservices architecture is not foreveryone, and adopting it will not suc-ceed unless the organization hasalready embraced several other modernsoftware development practices andtechnologies:n Agile development: In a microser-vices architecture, each service is single

purpose, and should therefore be devel-oped by a fairly small team. If a problemis encountered, a fix can be quicklydeployed for that service alone, withoutaffecting the stability of other services.That’s the essence of agile development:small teams delivering frequent releasesin short iterative cycles. If your develop-ment process is still waterfall or “fragile”(faux agile, i.e., waterfall disguised bydaily standups), you won’t be able todevelop or maintain microservices atthe required pace.n Automated testing: To maintainand upgrade a microservice quickly, youneed to be able to quickly assess theimpact of a change. Multiply this by hun-dreds or thousands of services, and youquickly realize that there aren’t enoughmanual testers in your company to keepup. The only solution is test automation,where unit tests and end-to-end systemtests are automatically run and theresults are automatically validated.n DevOps: A microservices architec-ture vastly increases the velocity andvolume of deployments, which canpresent severe operational challenges.To solve this problem, you’ll need tobreak down the walls that separatedevelopment from operations. That isprecisely the goal of DevOps, which isthe practice of operations and develop-ment engineers participating togetherin the entire product lifecycle, fromdesign through development to pro-duction support. The core values ofDevOps are summed up by theacronym CAMS, which stands for:

Culture: DevOps is first and fore-most about breaking down the barriersbetween development and operations,fostering a safe environment for inno-vation and productivity and creating aculture of mutual respect and coopera-tion. This value may sound “soft”, but itis the most important (as Peter Druckeronce observed, “Culture eats strategyfor breakfast”), and it is the most diffi-cult to implement, because there are noshortcuts or workarounds for the hardwork of changing people’s attitudes.

Automation: To develop high qual-ity microservices at scale, you’ll need toperform continuous integration, whereregression tests are automatically runeach time a piece of software changes.To deploy at scale, you must eliminatethe possibility of human error byautomating the deployment process, tothe point where your infrastructure iscode, i.e., recipes that that have beenproven correct and can be run ondemand across a myriad of machines.

Measurement: You need to capturemetrics about each stage of the devel-opment and deployment process, andanalyze those metrics in order to createan objective, blameless path ofimprovement.

Sharing: A key to the success ofDevOps in any organization is sharingthe tools, techniques and lessons acrossgroups, so that duplicate work can beeliminated and so that teams can makenew mistakes instead of constantlyrepeating other teams’ mistakes.n Docker: You’ll need a well-defined,


Moshe Kranc is CTO at NessDigital Engineering

Thank You forNot Adopting MicroservicesUnless you’ve already adopted modern development practices



right microservice given the URL thatthe traffic is addressed to. Here we canapply HAProxy, NGINX or AWS Appli-cation Load Balancing (ALB).

AWS ALB was introduced in Augustand in the short time it’s been available,a debate has emerged as to which tool isbest for application load balancing. Twokey questions you might ask to makethe right decision are, how manymicroservices do you plan to have andhow complex do you want your routingmechanism to be.9. What tool will be used for secrets?With the number of microservices in agiven application expected to increaseover time, and modern applicationsrelying more and more on SaaS extend-ed solutions, security simultaneouslybecomes really important and more dif-ficult to manage. For microservices tocommunicate with each other, they typ-ically rely on certificates and API keysto authenticate themselves with the tar-get service. These API keys, also knownas secrets, need to be managed securelyand carefully. As they proliferate, tradi-tional solutions, such as manually inter-jecting at time of deployment, don’twork. There are frankly just too manysecrets to manage, and microservicesrequire automation.

Organizations need to settle on anautomated way to get secrets to con-tainers that need them. A few potentialsolutions include: • In-house solution built for saving

secrets in encrypted storage, decrypt-ing them on the fly and and injectingthem inside the containers usingenvironment variables.

• AWS IAM rules which can interjectAmazon API keys. However, this solu-tion is limited to Amazon API keysand can only be used to access secretsstored in other Amazon services.

• HashiCorp Vault uses automation toeffectively handle both dynamic andstatic secrets. Vault is a very extensivesolution with several features unavail-able in other solutions and we arefinding it to be a more and more pop-ular choice going forward.Your answer to this technology ques-

tion depends on how many secrets you

have; how you expect that number togrow; your security and complianceneeds; and how willing you are tochange your application code to facili-tate secret handling.10. Where will SSL be terminated?One question that arises frequently,especially around microservices thatservice web traffic is: where should SSLbe terminated? Typical design factorsto consider include your security andcompliance requirements. Typicaloptions are at the application or net-work load balancer, for example termi-nating them at AWS ELB or ALB. Asecond option is to terminate SSL at anintermediate layer such a NGINX, or atthe application container itself.

Certain compliance initiatives, likeHIPAA, require that all traffic beencrypted. Thus, even if you decrypt atthe load balancer, it needs to be re-encrypted before it is sent to containersrunning the application. On the flipside, the advantage of terminating atthe load balancer is that you have a cen-tral place for handling SSL certificates.And fewer things have to be touchedwhen an SSL certificate expires orneeds to be rotated.

Elements to consider as you make adesign decision include your specificcompliance and security requirements;the ability of your applications toencrypt and decrypt data; and your con-tainer orchestration platform as somehave the ability to encrypt data seam-lessly. The combination of all the aboveshould be the basis for your SSL termi-nation decision.

While all these design and technolo-gy points may feel overwhelming, making the right choices will have long-term implications to your organization’ssuccess with its microservices architec-ture. Like painting a house, more thanhalf the work is in the preparation.From choosing the right primer toproperly taping the wall, setting theright foundation and process bound-aries are of significant importance inplanning Docker-based microservices.Don’t short-change your preparatoryprocess and you’ll end up with an end-product that delivers on your organiza-tion’s most critical microservices goals. z

< continued from page 36consistent, self-contained environmentto run each of your microservices. Themost popular environment today isDocker, a Linux-based container whichprovides fast provisioning and low per-formance overhead. Implementingsome microservices will require runningseveral cooperating containers, e.g., onecontainer for a NoSQL database andanother container that generates dataand stores it in the NOSQL database.To manage these container swarms,you’ll need to master a toolchain thatincludes tools like Kubernetes (fororchestration), Spring Boot (for deploy-ment), Consul (for service discovery)and Hystrix (for failure recovery).n Cloud computing: The cloud is theideal environment for deploying amicroservices architecture, because thecloud provides the needed scalabilityand automation, with higher reliabilitythan on-premise infrastructure. But,you’ll need to choose a cloud providerand then familiarize yourself with theirtools and interfaces for deploying andmonitoring cloud-based applications.

If your organization is comfortablewith all or most of these practices andtechnologies, then you are ready to enterthe world of microservices architecturesand reap the benefits. If not, then yourorganization is not ready for microser-vices, and the pain of attempting such aproject will probably far outweigh thebenefits. You would be best served in theshort term by adopting a more traditionalnon-monolithic architecture that is bet-ter suited to your development culture,e.g., a Service Oriented Architecture(SOA) based on coarse-grain services.

For the longer term, you’ll need to upyour game, because the rate at whichapplications are released and updatedwill only increase over time, and you’llneed to embrace modern software devel-opment practices that support the newpace of business. Unfortunately, thereare no short cuts to cultural change. Gethelp from a partner who can guide yourorganization through adoption of Agileand DevOps practices, so that you canthen painlessly benefit from microser-vices architecture and whatever comesafter that in the ever-evolving world ofsoftware engineering best practices. z



http://www.collab.net


Velocity. Quality. Traceability.Scaleability.For companies, it’s these char-

acteristics that form the definition of“DevOps.” But it doesn’t stop there;doing DevOps right also means consid-ering the processes, the culture, andthe tools to deliver software and solu-tions for today’s value-hungry cus-tomers.

And it still doesn’t stop there,because DevOps is all about havingend-to-end perspective while embrac-ing continuous integration and continu-ous delivery. As an organizationmatures, it will most likely includesome modern CI and CD practices, butto do this right, but it’s going to takemore than just a few automated tasks tomake development easier.

To be truly end-to-end, with visibil-ity and constant feedback, the businessneeds to realize it’s not about one singletool, it’s about a chain of solutions and


Buyers Guide

Getting an end-to-endperspective DevOps requires global view to successfully implement continuous integration, delivery

BY MADISON MOORE

SDT01 page 41,42,45,46,49,50_Layout 1 6/15/17 4:36 PM Page 41


how to pull them all together into agiant tool that operates as one. This iswhat will fuel a continuous integrationand continuous delivery workflow, withfeedback, visibility, and the ability todeliver services to customers quickly.

Accepting DevOps Regardless of the industry, change isalways difficult, whether it’s changing aprocess, culture, the methodology, or aparadigm. It’s natural, and often expect-ed, to feel a sense of resistance or hesi-tance.

Getting teams to accept change,especially when it comes to workflow orculture, is part of the overall challengeof DevOps. According to Rod Cope,CTO of Rogue Wave, some teams willembrace it quickly, agreeing that it isthe key to efficiency and scalability.Those that resist, however, fear thatautomation will make them obsolete,and so they inject manual steps alongthe way to make sure they seem valuedin the organization, even if this doesslow things down, said Cope.

“It can be challenging to get thosepeople over the hump to say well,there’s always more work that can bedone so it’s not like we are going to needless people,” said Cope. “We are justgoing to move faster.” Instead, theseautomation tools allow organizations tobe more flexible, keep up with rapidlychanging technology, and it will let theorganization move faster, he added.

The reality is, DevOps is not aboutthe tools. The tools are the enabler ofthe “holy grail” of what companies aretrying to get to, which is delivering moreto customers faster with better quality,said Nicole Bryan, vice president ofproduct management at Tasktop.DevOps is trying to broaden the scopeof the conversation to include this cul-ture shift, which Bryan believes will takesome time. However, she does see theconversation becoming inclusive ofmore than just the development side.

“Ten years ago people didn’t realizedelivering a software product was aboutmore than just the code,” said Bryan.“And I think people do realize thatnow.”

Don’t be a ‘fool with a tool’Unlike the days of waterfall develop-ment, continuous delivery refers to theprocess of improving the overall releasepipeline, which includes deploymentautomation, provisioning, testing, andcontinuous integration. Developerchanges on a shared repository aremerged daily, sometimes several timesa day, giving teams the ability to workfast and stay involved all the time.

According to Thomas Hooker, vicepresident of marketing at CollabNet,the very ideas of continuous integrationand continuous delivery flow very natu-

rally into DevOps. The people and theprocesses are working rapidly together,and then the proper tooling supportsthese interactions, he said.

He said the right tool can have atremendous impact in the process fornot just the developers, but the QA testteam, the build team, and everyone elseinvolved. It’s these tools that give teamsthe visibility, the scalability, and thetraceability that is needed to delivervalue to the customer.

“At the end of the day we work toserve the customer,” said Hooker. “The



Just like DevOps is about more than just tools and processes, continuous integrationand continuous delivery is more than just automating code and deploying changes. CIand CD can make development easier, but it’s going to take more than an automationtool to do so.

Getting CI and CD to make development easier means developers are checkingcode daily, and when the CI/CD pipeline fails, then new work stops, said John Jeremi-ah, IT and software marketing leader at HPE. Diagnosing and correcting the buildbecomes the top priority in this scenario, and either the code is fixed or it’s pulledfrom the repository to correct the build.

“Disruptive? Perhaps, but in the long run, quality improves, velocity improves, asdoes productivity,” said Jeremiah. “ Consider how security improves when every buildis reviewed, scanned and validated from an app security perspective. Letting the buildgo red and not immediately correcting the issues is the recipe for a mess.”

Having this discipline to correct build issues before introducing new changes is thekey to making development easier with CI and CD, said Jeremiah. And of course,automation is a huge benefit of CI and CD. Automating all the individual pieces makesit easier for the developer, and he or she can focus on what they need to develop,without having to worry about the external pieces, said Stephen Feloney, vice presi-dent of products for application delivery at CA Technologies.

Also, a good CD platform will remove unnecessary manual tasks and error-proneprocesses, which means developers can focus on functionality instead of non-codingtasks, according to vice president of products at XebiaLabs, Tim Buntel.

“Things like provisioning, deployment, compliance, approvals, status reports: all ofthat still happens but developers don’t have to deal with them because they’re auto-mated and standardized,” said Buntel.

—Madison Moore

How does continuous delivery actuallymake development easier?


mara

Highlight

mara

Highlight

mara

Highlight

http://www.roguewave.com/cd

http://www.tasktop.com


Deliver quality applications rapidly, and at enterprise scale. Manage tests with an integrated ALM toolchain built for waterfall and Agile application development. Grow from defining and managing work items tracking, to optimizing program and portfolio. Project Agile is not Enterprise Agile. Discover the New.

Visit saas.hpe.com/software/alm-octane

Does balancing speed, quality and scale feel like rocket science?

Blast off with HPE’s Application Lifecycle Management.


http://www.saas.hpe.com/software/alm-octane


http://www.ca.com/continuous-testing


way we best serve the customer is toprovide them value, and when we pro-vide them real consumable value, theywill buy more from us — not becausethey have to, but because they want to.”

Simply using a continuous integra-tion tool or an automation tool does notmean you are doing DevOps or doingCI. There is no set of tools that compa-nies can buy that will give themDevOps. Continuous delivery supportsthe promise of DevOps, according toTim Buntel, vice president of productsat XebiaLabs, but it’s the companiesthat “must be willing to change theirorganizational structures, communica-tions and processes to be suc-cessful with DevOps,” he said.

According to John Jeremi-ah, IT and software marketingleader at HPE, continuous deliv-ery tools are insufficient if the com-pany doesn’t have the discipline tocommit frequently, build often, testoften, and correct build issues.

“The tools help immensely, but with-out the right discipline, you can becomea fool with a tool,” said Jeremiah.

Companies can easily become a“fool with a tool” if they think that thereis a one-size-fits-all solution. That does-n’t exist, and frankly, it never did exist,said Bryan. For each team, there isoften a mix of tools automating theircontinuous integration or their continu-ous delivery workflow.

Focus on integrationWhile there is no tool to rule them all,leaders in the continuous delivery andcontinuous integration tool space saythat there is one important aspect that agood tool chain must include, and that isthe ability to integrate third-party tools.

That means, according to StephenFeloney, vice president of products forapplication delivery at CA Technolo-gies, if you have a continuous integra-tion tool or orchestration tool, it needsto be able to integrate with say, Jenkins,and it should integrate with other third-party tools across the board, he said.

At XebiaLabs, Buntel said that inte-gration is the most important part of asolid continuous delivery tool. Small

teams have it easier, but the reality ofmost organizations is there are severaldifferent tech stacks involved in build-ing software — lots of different envi-ronments and a mix of skills and lan-guages can easily complicate things.

The best platform will be flexibleenough to accommodate all these dif-ferent user types, these features, theseintegration points, as well as differentparts of the organization, he said.

Cope agrees, and said that integra-tion is key for whatever tool is chosen.The tool should be able to work withthe other tools in an environment,whether that’s Puppet, Docker, Jenkins,or other tools in the marketplace,

including open-source tools. “You are going to use a

bunch, and I like to say open-source islike potato chips, you can’t have justone,” said Cope. “You’re going to endup with a whole bag [of tools] at onepoint, so make sure they place nicelywith each other.”

And, flexibility is crucial here, saidCope. Every company has differentpackaging processes for their software, sopreserve your choices and be modular.The only way to adapt, he said, is to notbe locked into one process or tool chain.

Scaling for the enterpriseWhen it comes to enterprises, whichare large and often complex in nature,organizations need to find a tool chainthat can actually scale. According toFeloney, scripting, coding and main-taining one application is different thanscaling for an enterprise of say, 600applications.

“From an enterprise point of view,you need something that can actuallyfully scale to that level and not have to

be hard coded or scripted for every oneof the applications,” said Feloney. “Andyou need something that followsdependencies.”

The scaling aspect of continuousdelivery and continuous integrationtools is not to be underestimated, saysTasktop’s Bryan. While there is the fearaspect of how a new tool might impacta job, the bigger concern should bewith scaling, she said. Rob Elves, Task-top’s senior director of product man-agement, said “when you scale up, it’snot just simply volume; it’s scaling outto include inbound and outboundprocesses that are coming into thatDevOps ‘infinity loop.’ ”

Scale is the biggest challenge Task-top hears from its customers, becauseit’s a “whole different ball game” tryingto do DevOps with 100 people versus10,000 people. This is where Bryan seesa lot of tools falling short. At a largeorganization, a tool needs to offer thatcollaboration, that communication, andhave all of the right information flow.

“It boils down to the ability to havethat value stream inclusive of full trace-ability with full streams going back andforth, real-time, all the time, the beable to satisfy that,” she said. “Scale isthe short answer.”

The need for speedToday’s software delivery business initia-tives are centered primarily around theidea of speed. The faster the continuousdelivery pipeline runs, the faster theteam gets feedback, which ultimatelymeans, the consumer is getting their val-ue, their product or service — faster.

According to HPE’s Jeremiah, enter-prise IT has been “chronically” unableto deliver at the speed of business. Partof this is due to a legacy of monolithicapplications, brittle architecture, andregulatory and compliance require-ments, he said. Regardless, software iscritical to digital-first businesses, and sothe software delivery teams are going to

There is no set of tools that companiescan buy that will give them DevOps.




mara

Highlight

mara

Highlight

mara

Highlight

http://www.xebialabs.com




Tim Buntel, vice president of

products at XebiaLabs:

For small companies orgreenfield projects whereyou’re starting from scratch

or you’re working with one team and anarrow set of technologies, implementingDevOps is fairly straightforward. Wherecompanies begin to run into difficulties iswhen they start to scale their DevOpsoperations across disparate teams andwhen projects start involving a lot differ-ent individuals with varied skill sets, dif-ferent technologies for building software,and deployment to many different envi-ronments.

The XebiaLabs DevOps Platform isbuilt with all these variations and thiscomplexity in mind. Our platform offersmany integration points that make it easyfor you to bring all your tools and teamsinto a Continuous Delivery pipeline, withfull visibility for both technical and non-technical team members. It allows you toproactively spot bottlenecks and poten-tial failures anywhere in the pipeline. Itlets you automate deployment to all kindsof environments, changes to workflows,compliance requirements, audit reports,and release management tasks.

Rod Cope, CTO of Rogue Wave:

Rogue Wave Software’s Con-tinuous Delivery Assessmentprovides a blueprint for con-tinuous delivery, helping

companies develop an effective roadmapto adopting automation in software deliv-ery processes, including architecting andoptimizing open source-powered buildand delivery pipelines. Open source toolslike Jenkins, Maven, Puppet, Chef, Docker,and Kubernetes drive continuous integra-tion and delivery in the modern enter-prise, but getting them all working welltogether can be difficult.

Each of these tools has many deploy-ment options, dozens of configuration set-tings, and a large range of performancetuning capabilities. Developers may getsomething running, but is it optimized? Willit break at scale? Is it secure? We guideteams to implement a successful continu-ous delivery process, adopt DevOps best

practices, and implement open source toolsfor maximum efficiency.

John Jeremiah, IT and soft-

ware marketing leader at HPE:

One of the problems facingenterprises is gaining visibili-ty across pipelines, integrat-

ing different tools to get a handle on over-all speed, quality and security. We’reinnovating on a number of CI/CD relatedinitiatives. I’ll highlight a couple. We’vebeen building ALM Octane as a platformto help enterprises manage and maintainvisibility into the health of their delivery.Octane helps to address the enterpriseneeds for mature CI/CD based delivery.

Because automated testing is so criti-cal, UFT Pro extends open source tools likeSelenium and makes it easier for teams tocreate and maintain their battery of auto-mated functional test scripts. These toolsare critical enablers in achieving speed,quality and security of faster app delivery,but they have to do the work.

Stephen Feloney, VP of prod-

ucts for application delivery

at CA Technologies:

Despite agile practices inmany companies today, there

are still bottlenecks within the releasecycle that severely impede the ability ofcompanies to deliver more value to theircustomers. In a recent survey by Comput-ing Research, DevOps pros stated that63% of delays were coming from Test/QAstage of the cycle.

Again and again, we hear from ourcustomers that they are missing the markon continuous delivery because they stillhave manual testing silos, manual releaseprocesses and a disconnected DevOpstoolchain.

It is evident that companies cannotachieve continuous delivery if they do notmodernize their testing and release prac-tices, and CA solutions are purpose-builtto solve these bottlenecks. We enableteams to generate test scripts fromrequirements, simulate test environmentsanywhere, access robust test data whenit's needed and execute performance andsecurity testing early and often. We

orchestrate continuous “everything”—development, testing, release andimprovement — with robust integrationsto open source, commercial and homegrown solutions across the DevOps tool-chain, including planning, CI, testing anddeployment tools.

Nicole Bryan, VP of product

management at Tasktop:

Many CI/CD tools focus onsolving “the right side” of theDevOps delivery pipeline i.e.

connecting releases to builds, automatingdeployments and monitoring of produc-tion application. Or in other words,improving the time to value from codecompletion to code in production.

Of equal importance is “the left side” ofthe pipeline i.e. efficiently turning cus-tomer requests into requirements, fea-tures, epics and stories. These two sidesmust be interlinked because the success ofthe left side influences the success of theright side (and vice versa). Both must beintegrated and communicating to create avalue stream that continuously builds soft-ware that meets customers’ needs.

An organization’s main problem isknowledge worker access to the rightinformation at the right time, and Tasktopsolves that. Integration relies on the real-time flow of project-critical information.

Thomas Hooker, vice president

of marketing at CollabNet:

When selecting a solid contin-uous integration or continu-ous delivery tool, one of the

things it has to have is visibility so man-agers and teams can see what is going onat all times. I think that is key, and youhave to be able to provide that visibilitybased on the persona of the user.

Our products give customers that visi-bility across the “whole forest,” so you getto see all the different work steps from acommon dashboard. This is what Collab-Net’s DevOps Lifecycle Manager providesmanagers with: a single view of all theValue Streams in their portfolio, as well asinsight into the ways in which these appli-cations contribute to the value the organ-ization delivers. z

What continuous integration/delivery challenges does your company’s solution solve?


SDT01 Full Page Ads_Layout 1 6/16/17 1:30 PM Page 48



n Atlassian: Atlassian offers cloud andon-premises versions of continuous deliv-ery tools. Bamboo is Atlassian’s on-premises option with first-class supportfor the “delivery” aspect of ContinuousDelivery, tying automated builds, testsand releases together in a single work-flow. It gives developers, testers, buildengineers, and systems administrators acommon space to work and share infor-mation while keeping sensitive operationslike production deploys locked down. Forcloud customers, Bitbucket Pipelinesoffers a modern continuous delivery serv-ice that’s built right into Atlassian’s ver-sion control system, Bitbucket Cloud.

n Chef: Chef Automate, the leader in Con-tinuous Automation, provides a platformthat enables you to build, deploy andmanage your infrastructure and applica-tions collaboratively. Chef Automateworks with Chef’s three open source proj-ects; Chef for infrastructure automation,Habitat for application automation, andInspec for compliance automation, as wellas associated tools. Chef Automate pro-vides commercial features on top of theopen-source projects that include end-to-end visibility across your entire fleet,tools to enable continuous compliance, aunified workflow to manage all change,enterprise grade support, and more.

n CloudBees: CloudBees is the hub ofenterprise Jenkins and DevOps, providingcompanies with smarter solutions forautomating software development anddelivery. CloudBees starts with Jenkins,the most trusted and widely-adopted con-tinuous delivery platform, and adds enter-prise-grade security, scalability, managea-bility and expert-level support. By makingthe software delivery process more pro-ductive, manageable and hassle-free,CloudBees puts companies on the fastestpath to transforming great ideas intogreat software and returning value to thebusiness more quickly.

n Dynatrace: Dynatrace provides theindustry’s only AI-powered applicationmonitoring that transcends the challengehuman beings struggle with to managecomplex, hyper-dynamic, web-scale appli-cations. Bridging the gap between enter-

prise and cloud, Dynatrace helps dev, test,operation and business teams light upapplications from the core with deepinsights and actionable data. We help com-panies mature existing enterprise process-es from CI to CD to DevOps, and bridge thegap from DevOps to hybrid-to-native NoOps.

n Electric Cloud: Electric Cloud is a leaderin enterprise Continuous Delivery andDevOps automation, helping organiza-tions deliver better software faster byautomating and accelerating build, testand deployment processes at scale. Indus-

n Comprised of the CA Application Test, CA Mobile Cloud, CA

Release Automation and CA Service Virtualization solutions, the CA Continuous

Delivery product portfolio addresses the wide range of capabilities (from pre-pro-duction to release) necessary to compete in today’s evolving digital economy. CA’shighly flexible, integrated solutions allow organizations to fully embrace the evolv-ing requirements of the software-driven business landscape, enabling rapid develop-ment, automated testing, and seamless release of mission-critical applications.

n CollabNet helps enterprises and government organizations developand deliver high-quality software at speed. CollabNet is a Best in Show winner in theapplication lifecycle management and development tools category of the SD Times100 for 14 consecutive years. CollabNetoffers innovative solutions, consulting, andAgile training services. The company proudly supports more than 10,000 customerswith 6 million users in 100 countries.

n HPE’s DevOps services and solutions focus on people, process and tool-chain aspects for adoption and implementing DevOps at large-scale enterprises.Continuous Delivery and Deployment are essential elements of HPE’s DevOps solu-tions, enabling Continuous Assessment of applications throughout the softwaredelivery cycle to deliver rapid and frequent application feedback to teams. Moreover,the DevOps solution helps IT operations support rapid application delivery (withoutany downtime) by supporting a Continuous Operations model.

n Rogue Wave helps thousands of global enterprise customers tacklethe hardest and most complex issues in building, connecting, and securing applica-tions. Since 1989, our platforms, tools, components, and support have been usedacross financial services, technology, healthcare, government, entertainment, andmanufacturing to deliver value and reduce risk. From API management, web andmobile, embeddable analytics, static and dynamic analysis to open source support,we have the software essentials to innovate with confidence.

n Transforming the way software is built and delivered, Tasktop’s uniquemodel-based integration paradigm unifies fragmented best-of-breed tools and auto-mates the flow of project-critical information across dozens of tools, hundreds ofprojects and thousands of practitioners. The ultimate collaboration solution forDevOps specialists and all other teams in the software lifecycle, Tasktop’s pioneeringValue Stream Integration technology provides organizations with unprecedentedvisibility and traceability into their value stream. Specialists are empowered, unnec-essary waste is eradicated, team effectiveness is enhanced, and DevOps and Agileinitiatives can be seamlessly scaled across organizations to ensure quality softwareis in production and delivering customer value at all times.

n XebiaLabs develops enterprise-scale Continuous Delivery and DevOpssoftware, providing companies with the visibility, automation and control they need todeliver software faster and with less risk. Global market leaders rely on XebiaLabs tomeet the increasing demand for accelerated and more reliable software releases.

CA Technologies:

CollabNet:

HPE:

Rogue Wave:

Tasktop:

XebiaLabs:

n FEATURED PROVIDERS n

A guide to continuous integration and delivery tools



mara

Underline

mara

Underline

mara

Underline

mara

Underline

mara

Underline

mara

Underline





http://www.roguewave.com/cd


have to embrace this pressure toincrease speed, quality and security.

Since things are moving fast throughthe system, teams need to manage thatentire release pipeline, said Hooker.Now that there are so many releaseshappening each day, a tool chain canhelp the team understand the realitiesof this new way of development.

While businesses want a tool to helpthem keep the continual stream ofreleases moving, both business leadersand technical people care about differ-ent metrics. Hooker said the technicalteams want to see technical informationon code and commits, but this meansnothing to the business owner, whomost likely wants to know whether anapplication is delivering value to thecustomer, he said.

“The ability to show what is going on

across your continuous delivery or con-tinuous integration toolchain in termsof what is now being called a valuestream, is very important,” said Hooker.“We can show how all this work takingplace is delivering value to the businessand value to the customer, and thoseare the big things.”

Feloney also hears customers sayingthey need to go faster, and his questionto them is always “why?” Once you drilldown, it’s obvious it’s not just aboutgoing faster, it’s about the company try-ing to achieve better business out-comes, he said.

What is inhibiting these companiesfrom going faster? Feloney said it’s amultitude of things, but first and fore-most, it’s the culture. Tools can certain-ly help customers “go faster,” but toolswill not fix the problems.

“I can provide customers with tools

all day long, but if they can’t fix theirculture it’s not going to help,” saidFeloney. “Most enterprises are so usedto doing waterfall and having thesehand offs and all these check boxes, andeveryone has a say about things beingreleased, there is all this bureaucracy,and that is what’s inhibiting this fasterdelivery.”

XebiaLabs’ Buntel said all of theircustomers’ initiatives aim for that goal— to deliver value to their customersfaster. And it’s these high-level businessgoals centered around digital transfor-mations that are translated down to ITand development, said Rogue Wave’sCope. Those teams need to be moreopen and move faster and becomemore flexible, he said, which gets trans-lated to having continuous integrationand continuous delivery streamlinedevelopment through production. z


try leaders like Cisco, E-Trade, Gap, GE,Qualcomm and SpaceX use ElectricCloud’s solutions to boost software pro-ductivity. The ElectricFlow DevOps

Release Automation Platform allowsteams of all sizes to automate deploy-ments and coordinate releases.

n JetBrains: TeamCity is a continuousintegration and deployment server thattakes moments to set up, shows yourbuild results on-the-fly, and works out ofthe box. It will make sure your softwaregets built, tested, and deployed, and youget notified about that appropriately, inany way you choose. TeamCity integrateswith all major development frameworks,version control systems, issue trackers,IDEs, and cloud services.

n Micro Focus: Micro Focus offers solu-tions to help businesses successfullyimplement DevOps and Continuous Deliv-ery. Silk Central allows users to gain con-trol, collaboration and traceability acrossall areas of software testing. Atlas isMicro Focus’ agile requirements deliveryplatform that enables development teamsto gather and define business require-ments in alignment with agile delivery.

StarTeam Agile provides support forScrum-based sprint planning, backlogmanagement and tracking.

n Microsoft: Visual Studio Team Services,

Microsoft’s cloud-hosted DevOps serviceoffers Git repositories; agile planning tools;complete build automation for Windows,Linux, Mac; cloud load testing; ContinuousIntegration and Continuous Deployment toWindows, Linux and Microsoft Azure; appli-cation analytics; and integration with third-party DevOps tools. Visual Studio TeamServices supports any development lan-guage, works seamlessly with Docker-based containers, and supports GVFSenabling massive scale for very large gitrepositories. It also integrates with VisualStudio and other popular code editors.

n Puppet: Puppet provides the leading ITautomation platform to deliver and oper-ate modern software. With Puppet, organ-izations know exactly what’s happeningacross all of their software, and get theautomation needed to drive changes withconfidence. More than 75% of the For-tune 100 rely on Puppet to adopt DevOpspractices, move to the cloud, ensure secu-rity and compliance, and deliver bettersoftware faster.

n Redgate Software: Including SQL Serverdatabases in Continuous Integration andContinuous Delivery, and stopping thembeing the bottleneck in the process, is themission at Redgate. Whether version con-trolling database code, including it in con-tinuous integration, or adding it to auto-mated deployments, the SQL Toolbeltfrom Redgate includes every tool neces-sary. Many, like ReadyRoll and SQL SourceControl, SQL Compare and DLM Automa-tion, integrate with and plug into thesame infrastructure already used forapplication development. Think Git orTeam Foundation Server, Jenkins orTeamCity, Octopus Deploy or Bamboo, forexample, and the database can be devel-oped alongside the application.

n TechExcel: DevSuite helps organizationsmanage and standardize development andreleases via agile development methodsand complete traceability. We understandthe importance of rapid deployment andare focused on helping companies makethe transition over to DevOps. To do this,we have partnered with many automationtools for testing and Continuous Integra-tion, such as Ranorex and Jenkins. Rightout of the box, DevSuite will include thesetechnologies. z


A guide to continuous integration and delivery tools< continued from page 49



http://www.roguewave.com/CD

SDT01 Full Page Ads_52tk_Layout 1 6/15/17 3:38 PM Page 52

http://www.hyphenate.io

http://www.lightningchart.com/do


Guest ViewBY CIARAN DYNES

Ciaran Dynes is vice president of

products for Talend. With the immense success of cloud platforms and

Software-as-a-Service (SaaS) models on onehand and the increase of subscription services on theother, it’s clear that software offered as a subscriptionis becoming the new standard. In fact, as early as2015, Gartner estimated that by 2020, 80 percent ofvendors would adopt a subscription model.

This change in the way companies use softwarehas frequently been said to reflect users’ demandfor flexibility. Indeed, companies are no longer will-ing to lay out a major investment to get equipped.They are looking to prioritize variability in theirspending based on usage and to ensure they benefitfrom the value of the software before making along-term commitment.

As business models rotate toward subscriptionservices, if technical developments don’t keeppace, a crucial piece of the puzzle will be missing.

Subscription servicesLet’s look at another essential aspect: the value soft-ware offers both IT and business users. This is wherethe real challenge is: The ability to provide IT man-agers with frequent releases that encapsulate currenttechnology innovation and customer demands. Forbusiness users, cloud solutions provide the ability tohave quick access to the computer resources andapps you need when you need them to help deliverqualitative results faster, cheaper and with higherquality in order to create a competitive advantage.

A perpetual license model also allows for peri-odic software updates. However, the rhythm ofthese updates and the frequency at which they areavailable to users cannot be compared with theongoing agility and innovation offered by providersof subscription services. This is not related to howtheir software is marketed, but rather to the ven-dor’s ability to establish a continuous cycle of inno-vation for its products.

Big Data and CloudThe continued growth in the use of big data and

cloud technologies is in and of itself a compellingproposition for continuous innovation. The speed atwhich these technologies become obsolete requiresusers to adapt at an unprecedented rate. Considerhow the platforms adopted by customers today canbecome obsolete in as little as 12-18 months —Spark replaced MapReduce in record time andSpark 2.0 is a revolution compared to Spark 1.6. It isessential for integration, processing and operatingsoftware vendors responsible for these massive vol-umes of data to get as close to the market as possi-ble, which means complying with key standardssuch as Hadoop, Spark and Apache BEAM. Not tomention they need to align themselves with theopen source communities defining them. In practi-cal terms, a company needs to anticipate the prod-uct roadmap needed to align with these innovativetechnologies and keep pace with customerdemands.

Open-source technologies— backed by a technicallyadept developer community —are particularly well-suited to acontinuous innovation model.Additionally, subscription services are a logical wayto embrace and foster a continuous innovation mod-el. Previously dominant or legacy software models,marked by “proprietary” software solutions and per-petual licensing, take 18 to 24 months to deliver newfeatures. Delivering new versions every 18 months issimply not viable for businesses.

To support the emergence of new data usesModern solutions for big data and cloud integra-tion must be at the front lines of technology inno-vation. Not only to address customers various andrapidly evolving challenges—including customerintimacy, business sustainability, agility andeconomies of scale—but also encourage the emer-gence of new data uses like streaming, real-timeinsights and self-service in order to gain a compet-itive advantage. z

Subscription models fuel innovation

Open-source technologies ...are particularly well-suited to a

continuous innovation model.

Statement of Ownership, Management, and Circulation for SD Times as required by 39 U.S.C. 3685 PS Form 3526; SD Times, publication number 0019-625, filed July 1, 2017, to publish 12 monthly issues each yearfor an annual subscription price of $179. The mailing address of the office of publication, the headquarters of business of David Lyman, President and Publisher; David Rubinstein, Vice President and Editor-in-Chief,is 225 Broadhollow Road, Suite 211, Melville, NY 11747. The owner is D2 Emerge LLC, 2 Roberts Lane, Newburyport, MA 01950. Known bondholders, mortgagees, and other security holders owning or holding 1%or more of total amount of bonds, mortgages, or other securities are David Lyman, 2 Roberts Lane, Newburyport, MA 01950, and David Rubinstein, 225 Broadhollow Road, Suite 211, Melville, NY 11747. The actualnumber of copies of the July 1, 2017 issue includes: total number of copies (net press run): 21,234; paid and/or requested outside-county mail subscriptions: 19,502; paid/requested in-county subscriptions: 0; salesthrough dealers, carriers, and other paid or requested distribution outside USPS: 0; requested copies distributed by other mail classes through the USPS: 0; total paid and/or requested circulation: 19,502; outside-county nonrequested copies stated on PS form 3541: 1,732; in-county nonrequested copies stated on PS form 3541: 0; nonrequested copies distributed through the USPS by other classes of mail: 0; nonrequestedcopies distributed outside the mail: 0; total non-requested distribution: 1,732; total distribution: 21,234; copies not distributed: 38; for a total of 21,272 copies. The percent of paid and/or requested circulation was91.84%. I certify that all information furnished on this form is true and complete. –David Lyman, Publisher.


SDT01 Full Page Ads_Layout 1 6/14/17 12:41 PM Page 54

http://www.gdpicture.com


Analyst ViewBY PETER THORNE

Peter Thorne is directorat analysis firm

Cambashi.If you can estimate — to the right order of mag-

nitude — the significant quantities in your soft-ware system, you will have a much better chance ofmaking good decisions about architecture, algo-rithms, data structures and deployment.

By the numbersThe single point I want to focus on is the role ofnumbers in decision making about software sys-tems. I’m talking about the numbers that definecharacteristics of the finished system — number ofconnected users/machines, scope, scale, speed,data throughput, acceptable down time and so on.

My belief is that you don’t need to know thesenumbers accurately, but it is absolutely vital toknow them to the right scale, the right order ofmagnitude — tens, hundreds, thousands, millions?To make good decisions at any level, one of themany things needed is to be able to identify whichnumbers are relevant, then estimate them. Fromnumber of records for a sorting algorithm to num-ber and complexity of moving objects in a videogame scene rendering system, the numbers guideyour decisions. Most of the time, being within afactor of 10 will be good enough. But most of thetime isn’t all the time — safety-critical softwareusually needs better estimates!

New rules for Internet of ThingsIOT is changing expectations, and introducing newrelevant quantities. One example is sensor data.

How much data should you expect when youhave five thousand sensors being scanned once persecond? Most people would say 5,000 data pointsper second. This is not wrong, but in practice it isan upper limit. Why? Because some sensors, or thefirst level of processing of sensor signals, reportchanges, not actual values. Imagine a lighting con-trol system in which software scans the on/offswitches, then decides what to do. On average, theswitch setting stays the same for long periods oftime. But everyone expects fast response of a lightto its switch. So the switch setting must be scanned10 or more times per second. But there is no needto report long sequences of ‘no change’, just reactto a few switch ‘events’ per day.

I recently saw an example for over 150 thousandsensors. The ‘typical’ scan rate was believed to be

once per second, so the rough estimate for data ratewas ‘one or two hundred thousand data points persecond.’ When measured, this system generatedbetween 850 and 1,350 events per second. So thereality — one or two thousand events per second —was two orders of magnitude, a factor of 100 times,smaller than the estimate.

The point is that a communications system opti-mized for 100K data points per second is probablysuboptimal for 1K data points per second in almostevery respect, including hardware required, store-and-forward queue sizes, data structures used forevent reporting, error detection/correction strategy,failover sequences and the interface to the manage-ment and reporting systems. Even choices of devel-opment environment, software technology and teststrategy may be wrong.

So the designers of this system need seriousconversations about what theyshould do. Do they need to han-dle a theoretical peak of 100times ‘normal’ capacity? Or per-haps the estimate was just wrong.

It’s not just speed!There are many quantities thatmay not be specified in requirements — for exam-ple, dataset size, range of screen resolutions, num-ber of simultaneous users, time before an inactiveuser is logged out, max and min update frequencyfor the code. Software engineers need to decidewhat numbers matter to the development methodand the software design. Sometimes it’s right to passthese questions back to the requirements author.

Dig inside the code, and other numbers becomesignificant — how many times will that code betriggered, is there a maximum iteration count forthat loop, how many ways should we index thatdata, and so on. This is almost always territorywhere there’s no way of passing the problem.

Ask, then answer or estimateSo in all cases, it’s right to question…how many,how big, how frequent, how long? Then think whatwould change if your estimate changed by an orderof magnitude. This will help you build insights intothe problem and the solution, and these insightswill lead you to build better software. z

Software engineers need todecide what numbers matterto the development method

and the software design.

Quantifying software quantities


SDT01 Full Page Ads_52,56TK_Layout 1 6/15/17 11:22 AM Page 56

http://www.syncfusion.com/SDunlimited

july 2017 vol. 2, issue 1 july 2017† vol. 2, issue 1† $9.95 † sdt01 cover_layout 1 6/16/17...

Documents