real-time business intelligence

Upload: boefbrak

Post on 08-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/6/2019 Real-Time Business Intelligence

    1/35

    Evolution In Information AccessReal-time Business Intelligence

  • 8/6/2019 Real-Time Business Intelligence

    2/35

    Evolving to real-timedata availability

    29%instantaneous

    44%daily

    24%weekly

    3%monthly

    Data: Gartner Survey of 540 organizations

    2006 2006 2002 2002

    55%daily

    11%instantaneous

    17%weekly

    17%monthly

  • 8/6/2019 Real-Time Business Intelligence

    3/35

    Real Time DataWarehouse DesignChallengesDr. Ralph KimballFounder Kimball Group

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    4/35

    Definition of Real Time

    Real Time is anythingthat is too fast for your current ETL

    When you reach the performance limitations of your current ETL architecture, you are forced tore-architect your system.

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    5/35

    Study the Alphabet Soup

    EII Enterprise Information Integration Direct access to source data in warehouse framework

    CTF Capture, Transform and Flow Database to database pipe

    SOA Service Oriented Architecture XML documents with WSDL payloads inside SOAP envelopes

    SETL Streaming ETL Record at a time transfer MBETL Micro Batch ETL

    Batch transfer every few minutes or seconds

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    6/35

    Real Time Requires More

    But when you re-architect your system, you mustinclude a whole set of other new data warehouserequirements Operational Interaction

    Extreme Integration Multi-modal data, especially text search Subtransactional data streams

    ComplianceSome of these conflict

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    7/35

    Operational Interaction

    Most extreme latency demand Always need the current truthNeed write privileges with locking

    Push, pull, and non-eventsGood application of EII Conflicts with integration and compliance

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    8/35

    Extreme Integration

    Local and global de-duplication in real time Need multi-valued back pointers to original IDsThe wider the integration,the more difficult to support real timeWorld is rapidly moving to SOA Service Oriented Architecture

    Abstract individual system details Compose coarse grain services from fine grain XML data transfer but relational targets

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    9/35

    Multi-modal data

    Especially text search Google has rewritten expectations

    for content and performanceAll of the web

    Deliver search result in less than 1 secondCan expect hybrid combinations of SQL and text search on same databases Fact Tables for Text Document Searching(Kimball article archive: www.kimballgroup.com) old Red Brick pattern index

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    10/35

    Subtransactional Data Streams

    Enormous volumes, up to 10 TB/day threatenconventional ETL and RDBMS modelsData streams allow sequential tracking of individualcustomers or productsWe barely understand sequential behavior analysis Most queries today look at flow at a single point

    Need to look at voyages through many points New design: Step dimension

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    11/35

    Snippet from Modeling Class:Design of a Step Dimension

    Design: Add a Step Dimension to a transaction fact table:

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    12/35

    Compliance

    Maintain the custody of data Prove later that data hasnt changed and

    that you computed the result correctly Show how information has been used

    Column-level encryptionHash-coded data sets Conflicts with operational interaction

    Where is the system of record? Compliance webinar (April) on Informatica web site

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    13/35

    How to Proceed(Building a RT Data Warehouse)

    Decompose the problem before implementing! Business Needs Compliance Data Profiling

    Security Data Integration Data Latency

    Archiving and Lineage End User Delivery Interfaces Available IT and End User Support Skills

    Legacy Licenses2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    14/35

    How to Proceed

    Choose an extensible architecture good at Simultaneous updating and querying Enormous data volumes, up to TB/day Multi-modal data search

    Compliance support Hardware independence

    Learn, learn, learn Sybase Education Services (www.sybase.com) Kimball University (www.kimballgroup.com)

    2005 Kimball Group / Kimball University All rights reserved.

  • 8/6/2019 Real-Time Business Intelligence

    15/35

    Evolution In Information AccessReal-time Business Intelligence

  • 8/6/2019 Real-Time Business Intelligence

    16/35

    Real-timeData Integration for Real-timeBusiness IntelligenceDavid LyleVice President, Product StrategyInformatica Corporation

    2005 Informatica Corporation

  • 8/6/2019 Real-Time Business Intelligence

    17/35

    2005 Informatica Corporation

    Real-time BI requires real-time DI

    Real-time Business Intelligence requires theintegration of data in real-time

    Kimball definition of real-time: Anything that is too fast for your current ETLYou need to re-architect your kitchen, but you still need a sink,oven, stove, fridge,

    I f i l i DI

  • 8/6/2019 Real-Time Business Intelligence

    18/35

    2005 Informatica Corporation

    Informatica real-time DIto the rescue

    Different appliances for your real-time kitchenEII virtual integration through query federation

    Capture, transform and flow = PowerCenter Real-time optionSOA PowerCenter Connect for Web servicesPowerExchange Change Data Capture options

    (Streaming ETL, Micro-batch ETL)PowerCenter Unstructured Data option

  • 8/6/2019 Real-Time Business Intelligence

    19/35

    2005 Informatica Corporation

    Capture, transform and flow

    Always-on sessionsPowerCenter Connect for MSMQ, IBM MQSeries, Tibco,

    webMethods, httpCan perform the extreme integration youre used to within thesame PowerCenter platform

  • 8/6/2019 Real-Time Business Intelligence

    20/35

    2005 Informatica Corporation

    Service-oriented architecture

    New way of thinkingNew programming language, too

    But SOA doesnt equal real-time data integrationSOA frameworks are enterprise stub-generators, leaving the dataintegration problem up to you!

    PowerCenter Connect for Web servicesBuild your data integration logic as Web services, backed up bymetadata and change management

  • 8/6/2019 Real-Time Business Intelligence

    21/35

    2005 Informatica Corporation

    Change data capture

    Approaches to dateAdd timestamps through reverse engineering

    User exits that interrupt transaction streamsAdd triggers that impact application performanceFull file compares to determine deltas

    Home-grown solutions technically challenging

    all disruptive, all programming-centric!

  • 8/6/2019 Real-Time Business Intelligence

    22/35

    2005 Informatica Corporation

    Value

    Simplify with one standard for all data types and latenciesExtend value of existinginvestments

    Accelerate project lifecyclesReduce solution developmentand maintenance costs

    Technology

    Complex source accessPatented changed data captureLatency: real-time, change, batchNon-invasive, service-orientedarchitectureCodeless, visual SQL drivenEnterprise scalability

    Informatica PowerExchange

    Unlock complex data on demand

  • 8/6/2019 Real-Time Business Intelligence

    23/35

    2005 Informatica Corporation

    Real-time analytics

    The reduction in data integration overhead may beits own justification for a real-time data warehouse

    Less data to move, reduce the load window, keepup with increasing data volumes, smaller systemand disk sizes requiredIts a win, win, win situation!

    And then if the business also has aBI requirement for real-time, thats obviously

    the most important justification of all!

  • 8/6/2019 Real-Time Business Intelligence

    24/35

    Evolution In Information AccessReal-time Business Intelligence

  • 8/6/2019 Real-Time Business Intelligence

    25/35

    Business Intelligenceon DemandInformation at the right time,right place, right priceKapi Attawar Senior Director Sybase, Inc.

    2005 Sybase, Inc.

  • 8/6/2019 Real-Time Business Intelligence

    26/35

    Sybase solutions

    2005 Sybase, Inc.

    ON-DEMANDMOBILITY SERVICES

    AvantGo

    Mobile Business Services AvantGo Premier Hosted PortalHosted Emails

    INFORMATIONANYWHERE

    SQL AnywhereM-Business AnywhereTM

    Mobile Link

    Unwired Accelerator Unwired Orchestrator

    Answers AnywhereTMMobile Sales FamilyPylonTM

    AfariaMach DesktopTM

    mFolioTM

    RFID Enterprise/RFID AnywhereTM

    HIGH-PERFORMANCEDATABASE

    ASE

    ENABLE BETTER DEVELOPMENTWorkSpace, PowerDesigner , PowerBuilder , PocketBuilder TM for client/server, Web and mobile

    INFORMATION MANAGEMENT (DATACENTER) INFORMATION MOBILITY (DE

    DATA

    KNOWLEDGE

    TAMING THEDATA EXPLOSION

    Heterogeneous Data Services

    Dynamic ArchiveMirror Activator TMDynamic ODS

    Sybase IQ Auditing

    Replication Server Real-Time Data Services

    Enterprise InformationIntegration (EII)

    Unstructured Data

  • 8/6/2019 Real-Time Business Intelligence

    27/35

    Business problemCritical data in silos and in

    different formsUsers need dynamic access todata in real time

    Solution: Avaki EII

    Cost-effective tool to developapplicationsSeamless access to disparatedata sources

    Provides customers with virtualviews of: Multiple data sources Federated data access SOAIntegration and query capabilities

    Business problemRapid growth of enterprise

    information outside structureddatabasesThis unstructured data comesin a variety of formats emails,videos, PowerPoint, etc

    Solution: ISDDProvides enhanced query andanalysis of both structured andunstructured dataAddresses both fixed andmobile sources

    Recent acquisitions

    2005 Sybase, Inc.

    Just-in-time

  • 8/6/2019 Real-Time Business Intelligence

    28/35

    Extract

    DataWarehouse

    Just in timeinformation architecture

    OperationalData Store

    ASE

    Analytics

    Business ObjectsCognos

    Microstrategy

    HyperionSPSSSAS

    Crystal ReportsMicrosoft Excel

    SAP BW

    Applix

    2005 Sybase, Inc.

    Just-in-time

  • 8/6/2019 Real-Time Business Intelligence

    29/35

    Transactions

    Access

    Avaki

    Extract

    DataWarehouse

    Just in timeinformation architecture

    OperationalData Store

    ASE

    Analytics

    Business ObjectsCognos

    Microstrategy

    HyperionSPSSSAS

    Crystal ReportsMicrosoft Excel

    SAP BW

    Applix

    The business opportunityDemand for real-time business reportingand analyticsImprove service to business by movingaway from batch reporting

    2005 Sybase, Inc.

    Wh d I ?

  • 8/6/2019 Real-Time Business Intelligence

    30/35

    What do I use?

    Sybase Dynamic ODS with InformaticaOperational system overload and higher operational costs causedby data and user growthCut costs reporting from OLTP systems is very expensiveImprove performance transactions currently adversely affectonline reportingNeed both current and historical data

    EII with Sybase AvakiOperational systems overload not the primary challengeOffload is not a good option too hard, data too dynamic, too muchfragmentation, mix of operational and OLAPReal-time accessibility with caching for performance and resourcemanagement

    Ideal for a grid environmentMoving to an SOA architecture 2005 Sybase, Inc.

    Real-time reporting

  • 8/6/2019 Real-Time Business Intelligence

    31/35

    Real time reportingDynamic ODS with Informatica

    ETL Tool (Informatica)

    Harvest data in near real-timeSource database agnosticHigh-speed data harvestingDisaster recovery capability

    Data on-demandwithout stressing theOLTP systemRisk-free approach,does not interfere withoperational systems

    Dynamic ODS (Sybase)

    Non-intrusive, risk freeData compression (up to 60%) for structuredand unstructured dataOptimized for fast-read access column-

    based, not rowStandard SQL access to data

    2005 Sybase, Inc.

    Real-time reporting

  • 8/6/2019 Real-Time Business Intelligence

    32/35

    ea t e epo t gEII with Avaki

    EII platform (Avaki)

    Real-time view of operational dataSource database agnosticHigh-speed data queries across distributedsystems

    Configurable caching for improved performanceand reduced operational impact

    Complex operationalenvironments

    Managed approachto operationaldata access

    Dynamic data transformationon-the-flyStandard SQL access to dataSupports updates and queries

    Service Oriented Architeture

    2005 Sybase, Inc.

    DatabaseEnvironment

    ETL Tool

    Data Sources

    Avaki Views

    Streamlining mainframes

  • 8/6/2019 Real-Time Business Intelligence

    33/35

    gand Teradata

    Dynamic ODS customer case studiesPicknPay (South Africa) Replaced 20 node Teradata with 8 CPU Xeon environment

    Major global credit card company (USA) Fraud detection investment of $4m = saving of tens of millions

    Telefonica (Europe) Single view of customer - investment $4m = savings of $15m in first year Mainframe, Oracle and Teradata co-existence

    HIRA (Korea) Enabled storage of five years of data on-line, improved query time from 4 days to 30 seconds Beat Teradata, DB2 and Oracle on performance and cost

    Avaki customer case studies

    Pfizer (USA) Daily sales dashboard real-time access to data from multiple data sources 600% improvement in analytic performance

    Pacific Capital Bank (USA) Risk management single view of customer and risk scoring

    Real-time access to data, transform and present in application-specific format

    2005 Sybase, Inc.

    Information at the right time,

  • 8/6/2019 Real-Time Business Intelligence

    34/35

    g ,right place, right price

    NeedsRelieve operational system overloadCut operational costsImprove transactional performanceProvide both current and historical data

    Solution: Sybase Dynamic ODS and Informatica

    NeedsOffload not a good optionAccess to operational system data in real-timeProvide a single virtual view of dataMergers and transitions

    Solution: EII with Sybase Avaki

    2005 Sybase, Inc.

  • 8/6/2019 Real-Time Business Intelligence

    35/35

    Evolution In Information AccessReal-time Business Intelligence