keynote- soa & beyond : future computing
DESCRIPTION
TRANSCRIPT
SOA & BEYOND
Future computing
Narendra Nathmal, Chief Architect, Cognizant
Technology Solutions
WSO2Con, Sep 15th , Colombo
AGENDA
Is SOA dead?
What will drive SOA in the future
Open source SOA capabilities, WSO2 experience
Future computing with SOA
High Performance Architecture
Semantics or Context driven Architecture
Event & time driven Architecture
Enterprise “Future”
2
WS
o2
Con
20
11
, Colo
mb
o
IS SOA DEAD?
Infamous article – enflamed passions – how could
it be?
Burton Group's Anne Thomas Manes on SOA
Can a concept die?
Was it misunderstood?
Did someone highjack SOA?
Is redemption possible?
If not will the world end?
3
WS
o2
Con
20
11
, Colo
mb
o
WHAT WILL DRIVE SOA IN FUTURE
Let’s get some facts straight
Customers are not gullible Fool them once … but try twice you risk losing them!
Domain experts are tech savvy – utilize their skills From eXcel spreadsheets to Business process
modeling, they know it all
Customers understand what it takes to build software They know waterfall does not work!
They know Services are important They can no longer afford to work in silos. There is
pressure to do things right
They know not every thing has to be purchased Customers know there is open source technology, they need
serious help here4
WS
o2
Con
20
11
, Colo
mb
o
OPEN SOURCE SOA CAPABILITIES
So you thing SOA is not possible with open
source … think again
DEMO
5
WS
o2
Con
20
11
, Colo
mb
o
FUTURE COMPUTING – IS SOA ENOUGH?
Customers need Customers!
And Customers are everywhere.
They will not call you … you need to connect with them?
How is that possible? Cloud, Mobile, Social networking …. ?
But these are just infrastructures or channels
SOA will have to couple itself with future computing styles High performance architecture
Intelligent (Semantic?) data processing
Event & time driven architectures
SOA style will have to address this Canonical modeling – design time tools & run time
transformations
Robust Infrastructure services
1:many deployment styles – EAI, Federation (Domain or Central) 6
WS
o2
Con
20
11
, Colo
mb
o
WHAT IS HIGH PERFORMANCE
ARCHITECTURE?
SLA and Response times is only part of the problem
The real challenge is how to create a efficient mix of high performance pipes within enterprise and robust integration (with better security) for external world interaction
Another challenge is integration of plethora of technology choices
Microsoft, Java (Legacy)
Deeply Entrenched (C++, PHP, Perl, Python)
Promising (Groovy, Erlang)
Exposing everything as a SOAP service is not the only acceptable solution often times 7
WS
o2
Con
20
11
, Colo
mb
o
WHO HAS BUILT HIGH PERFORMANCE
INFRASTRUCTURES?
eBAY, Facebook, Google etc
Thrift architecture
High performance serialization
Multi language support
Client & Server size code generation
IDL to define message structures & operations
8
WS
o2
Con
20
11
, Colo
mb
o
THRIFT – OPEN SOURCE HIGH
PERFORMANCE PLATFORM
9
WS
o2
Con
20
11
, Colo
mb
o
SAMPLE ADOPTION
Manufacturer Shop floor
Real-time
Applications
.NET
Enterprise portal
PHP
Application
Farms
JAVA
Shop floorShop floor
dashboard Enterprise
DWEIS
High speed
serialization
High speed
serialization
Shop floor
dashboardShop floor
10
WS
o2
Con
20
11
, Colo
mb
o
SEMANTICS POWERED ARCHITECTURE
Can Semantics be described as domain specific ontology's which machines can process?
What is Ontology?
Why is it important
Where it is most applicable
SOA & Semantics – How can they benefit from each other
11
WS
o2
Con
20
11
, Colo
mb
o
Immunodeficiency
syndrome
T-Cell
imm.
AIDS
Cancer Ontology
Use the ontology to retrieve the
diagnosis when
'Immunodeficiency Syndrome‘
is the condition
“Book me a holiday next weekend somewhere warm, not too far
away, and where they speak French or English”• Has meaning or semantic(s) – namely a person wants to go on a holiday
WEAVING SEMANTICS INTO A SERVICE
Service
Financial
Context
Non-functional
Context
Functional
Context
Cost to use
the service
Average
response
times
Service
Capabilities
Service
Parameter
Data
SemanticsService
Definition
Uptime /
Downtime
Security
Policies
Metering
Model
Chargeback
Model
Technical
Context
Service
access data
store
Service
composition
scenarios
12
WS
o2
Con
20
11
, Colo
mb
o
HIGH LEVEL VIEW
Providers Consumers
Service
Registry
Semantic SOA Framework
Service Bus
Service Service Service
UI Tooling
Service Metadata Ontology Artifacts
FunctionalNon-
functionalFinancial
Semantic SOA Engine
Goals Processor Mediation Selection
DiscoveryReasoner Composition
Service Description
Technical
13
WS
o2
Con
20
11
, Colo
mb
o
SAMPLE ADOPTION
Telecom operations support systems
Enterprise
Operations
systems support
S
E
R
V
I
C
E
S
Semantic
Service
Discovery
RDF based
vocabulary
OSS1 OSS2 OSS3 OSS4
WS-*
14
WS
o2
Con
20
11
, Colo
mb
o
External
providers
Medical: Drug
detection
SAMPLE SCENARIOS
Ability to process prescriptions to medical codes for
insurance payoutsMedical:
Medical coding
Provide contextual help to a defense lawyer working on a complex litigation on
similar cases, its outcome, hurdles, prosecution strategy etc
Legal:
Contextual help
Detect drug or disease based on name, chemical
composition or symptoms or other related characteristics
Discover services based on customer type, location, time sensitive semantics
to create a personalized page
Retail:
Service
composition
15
WS
o2
Con
20
11
, Colo
mb
o
EVENT AND TIME DRIVEN ARCHITECTURE
What is an Event?
What is a complex event?
Events in real world – some examples
Event and time – what’s the connection?
What class of applications can benefit the most
16
WS
o2
Con
20
11
, Colo
mb
o
SAMPLE SCENARIOS
Discover fraudulent activity by detecting patterns among events.: Single ID
card used twice to enter subway in less than 5 secs. := alert security for
piggybacking
Larger than average deposit made by the customer. Check if this is in
preparation of a larger purchase and the if the customer is looking for a
loan or investment vehicle
Tracks transactions that are traded at a level exceeding
a pre-determined % threshold away from pre-defined
benchmark
Associating multiple transfers that individually might not seem
suspicious.
Identifying fraud with credit card: purchase with same credit card
happened in London and Hong Kong within 6 hours (Minimum travel time
between London and Hong Kong is more than 10 hours)
Items coming and leaving warehouse can be monitored using RFID tags
Automatically order is placed to the supplier when items reaches below the
threshold
Financial:
Market
Surveillance
Financial:
Market Retail
Banking
Financial:
Market Fraud
Detection
Financial:
Market Anti
Money
Laundering
Retail :
Automated
Supply Chain
Transportation:
Security and
Fraud Detection
Reduce false positive alarms: When 15 alarms are received within any 5 second
window, but less than 5 similar alarms detected within 30 seconds, then do
nothing
Energy and
Telecommunicatio
ns: Alarm
Correlation
Automate stock trading based on market movement: if, within any 20 second
window, StockB rises by more than 2% and StockA does not, then automatically
buy StockA.
Financial:
Algorithmic
Trading 17
WS
o2
Con
20
11
, Colo
mb
o
SAMPLE ADOPTION
Push based paradigm
….
….
Event Normalization
BAM ChartsEvent
Publishing
Event
MetadataEQL
Event
Persistence
….
….
….
….
Downstream
systems18
WS
o2
Con
20
11
, Colo
mb
o
WHY IS DATA MODELING IMPORTANT?
Canonical data model is not new but there is
renewed interest in it
Is it possible to model data at rest and data in
motion and keep everything in sync?
What technology choices are available here
E2E modeling & integration – More useful for EAI
style of integration
Focus on modeling but customize integration for
better performance – here SOA can help
19
WS
o2
Con
20
11
, Colo
mb
o
SAMPLE ADOPTION
Insurance domain (Model to ACORD standard)
20
WS
o2
Con
20
11
, Colo
mb
o
This is what real life will demand!
This is real today!
Governance
Services
ENTERPRISE “SOA-IN-FUTURE”
SOA
Tool set
Semantics
Event &
Time
High
performance
Web Services
Integration Services
Business process Services
Rules Services
Data
Mod
eli
ng
Sta
nd
ard
iza
tio
n
Discoverable services Push based Services Superfast Services
WS
o2
Con
20
11
, Colo
mb
o
21
Thank you!
22
WS
o2
Con
20
11
, Colo
mb
o