a2: enterprise business intelligence with open source...
TRANSCRIPT
A2: Enterprise Business Intelligence with Open Source
DRAFT
Speakers:Christopher Lavigne, Partner, Breadboard BI.
Zachary Zeus, Principal, BizCubed.
Agenda
Introductions Overview Components of Enterprise Business Intelligence (BI) Some Common Open Source BI Tools Open Source BI Demonstration Open Source Tool Implementations Conclusions Questions & Answers
Introductions
Chris Lavigne Partner, Breadboard BI Architect for packaged BI
modules for Open Source Platforms.
External Pentaho Partner. External MySQL Partner. 10+ Years BI experience,
report writer to director.
Zachary Zeus Principal, BizCubed Led Implementations of
several Open Source BI implementations.
External Breadboard BI Partner.
External Pentaho Partner. 10+ Years BI experience.
Overview - Terms
Is Business Intelligence Something New?− Data for Decision-Making Utilized by Successful Scientists,
Generals, Politicians, and Business People for thousands of years.
Ever Changing, Flashy Terms & Acronyms are Confusing. New Terms Arise for Maximum Marketing Effect.
− DSS, EIS, DM, DW, Packaged Analytics, BI, OSBI− ETL, ELT, ETLT, EAI, EII− DB, DS, RDBMS, ROLAP, HOLAP, MOLAP, OLAP, MDBMS.− Dashboard, Report, Cube.
Overview – Enterprise BI System Basics
Simplify the Discussion – Know Yourself and Seek Improvement.
Understand How to Measure Enterprise Performance, for example via sales, enrollment, event attendance, battlefield metrics, etc.
Gather & Rationalize Data from Multiple Sources Inside & Outside the Enterprise.
Store Data in Structures Optimized for Presentation. Disseminate Data to Users Across Multiple Departments. Tool Selection based on Enterprise Requirements,
not buzz words, golf balls, sales pitch appeal.
Gather & Rationalize Data from Multiple Sources
Common Enterprise Data Souces:− Enterprise Resource Planning (ERP) Solutions
Like Compiere, JDE, PeopleSoft, Oracle, SAP, SugarCRM.
− Custom-Built Systems for Processes Like Orders, Invoices, Inventory, etc.
− Non-System Internal Sources Like Spreadsheets.
− External data Like Credit Reporting Agencies, Industry Benchmark Data, Currency Conversion Rates, etc.
Gather & Rationalize Data from Multiple Sources
Common Methods− Hand-Code
Using Perl, JAVA, Transactional SQL, etc.
− Proprietary Tools Like Ab Initio, Ascential, Informatica, OWB, etc.
− Open Source Tools Like CloverETL, KETL, Kettle (Pentaho Data Integration),Talend Open Studio, etc.
Common Problems− Undocumented Sources− Changing Sources− Large Data Volumes− Maintainability of Code− Source Formats
Gather & Rationalize Data from Multiple Sources
Demo
Store Data in Structures Optimized for Presentation
Most Important – Use Proper Aggregation, Denormalization, and Indexing techniques.
Common Data Storage Methods− Relational Databases
Proprietary - DB2, Oracle, SQLServer, etc. Open Source - MySQL, PostgreSQL, etc.
− Multi-Dimensional Databases Proprietary - Essbase, Analysis Services, PowerPlay, etc. Open Source - Mondrian, Palo, etc.
Can the Database Handle Your Volumes? Is it fast enough?
Store Data in Structures Optimized for Presentation
Demo
Disseminate Data to Users
Common Methods to Share Information Across the Enterprise Using a Web Browser:
− Reports.− Cubes.− Dashboards.
Tools− Proprietary
Business Objects, Cognos, MicroStrategy, Oracle, etc.− Open Source
Actuate, JasperSoft, Pentaho, etc.
Disseminate Data to Users
Demo
Disseminate Data to Users
Demo
Disseminate Data to Users
Demo
Custom-Built or Pre-Built BI Systems
Custom Built by Internal IT or External Consulting Resources.
− System Designed to Requirements.− Time and Labor Costs.
Pre-Built.− Lower Time and Labor Costs.− Inherit KPIs/Best Practices from Previous Customers.− Still Requires Some Customization and Enhancements to Meet BI
Requirements.− Extra Database Objects That Are Not Required.
ProprietaryCognos, Business Objects, Oracle, Microstrategy, etc.
Open SourceBreadboard BI (using Pentaho).
Considerations
They may meet your enterprise requirements, but...− They may fall short in a head-to-head comparison with
proprietary tools...but do they meet your requirements?− Understand what functionality is included in
“community” v. “paid” open source products.− They can be buggy, try them before you believe
functionality claims. − Pay for support for production applications.− Resources can be hard to find.− The skills required to build an Enterprise BI Solution are
not “Proprietary” or “Open Source”, resources may not be significantly less expensive.
Implementations Client Considerations for Open Source Tools vs Proprietary
Tools.● Internal Adoption of Open Source
● Is the technical team familiar with managing Open Source projects?
● Will the business view the project through an open source lens?
● Consider not discussing the fact that it is open source.● Do the timeframes and client requirements mandate
consideration of proprietary tool?● Rate of Implementation (Availability of Consultants)● Value Proposition to Client● Technical Needs
BI is BI is BI
Proprietary tools provide some value.− Shortcuts, Wizards, and Templates − May reduce implementation times, but at a cost.
Underlying concepts, work and approaches are consistent.− Star Schemas - Slowly changing
dimensions− Data Integration - Aggregates, sums, averages− OLAP, SQL - Report design &
distribution Tools aren't a substitute for well designed architecture and
experienced team.
Implementation Dimensions
Anywhere from 6 weeks to 4 months.− Driven by clients engagement, data size, and
complexity of load processes. Servers
− Can be small 1 GB RAM, 2Ghz processor.− or med 8GB, Dual Quad Core processors.
Don't skimp on hardware – data transformations, analysis and reporting are resource hungry applications.
Time and money spent on analysts, business people and consultants will far outweigh the costs of a decent server.
Our Clients
Common Themes− Family Owned
Want the functionality of bigger organizations. But the costs are felt personally.
− IT Industry Start-ups – can't afford the big end of town.
− Usually driven by a single “leader” in the organisation. A person that has had success with “Open” projects.
Some of Our Clients IMAN Health Insurance Toowoomba City Council InTouch Direct DeBortoli Wines Fallshaw Wheels and Casters MooterMedia Tickit Systems Digital Armour Polonious Systems
Successes
Able to build and deliver projects that are more scalable, robust and performant at a lower cost.
− Can meet clients needs effectively without expensive tools.
− Can provide early results with relatively small client commitment.
− Continue to refine project and build on early wins.− Able to deliver high business value in a competitive
manner.
Challenges
Clients will say “I have this excel report...”− Most times you can produce something similar, but at
times it takes some creative thinking to fit client requirements to tool functionality.
There tend to be use cases that are not covered by the tools functionality.
− Explain to clients that it can be built, but at a cost. Try to get them to fund new tool development.
Clients expectations of tools (both BI & schema) don't account for the complexity of their internal processes.
Building a BI implementation requires focus of clients team.
What are clients looking for when they contact us
Sex and Sizzle− BI components may be the only thing that senior
managers see about the entire IT infrastructure.− This is by far the thing clients want the most.
A quick fix to internal data problems.− BI is not a silver bullet...
Free BI implementation.− Don't understand the costs and consultation necessary
to implement things correctly.
Conclusion
Pre-Qualify potential clients or your own organization.− Make sure that there is an internal leader that can help
drive the process – it may need to be you.− Make sure that there is a budget.− Evaluate data and client expectations around data
problems. Deliver something “sexy” in first deliverable and every
subsequent deliverable.− The guys on top are the ones writing the checks.
Questions?
Chris [email protected]
http://www.breadboardbi.com/
Zachary [email protected]
http://www.bizcubed.com.au/