Sun microsystems bi infrastructure
- 1. SUN BUSINESS INTELLIGENCE DATA WAREHOUSE (BIDW) SOLUTIONS Infrastructure Made Easy Jim McManus Managing Consultant - BI/DW Sun Microsystems, Inc.
- 2. Agenda • Trends and typical requirements • BI Platform (SW and Infrastructure) • Data Integration • Data Warehouse • BI Reporting and Analytics • BI Access Layer • Summary
- 3. Trends in BI and DW • User population Everything is Increasing! • Security • Volume of queries, reports, analysis • Enormous diversity of data • Mass of customized data • Regulatory requirements
- 4. BIDW Requirements Flexibility Long-Term Foundation Scalability Cost-Effective BIDW Performance Quick ROI High Availability Enterprise Security
- 5. Smarter Services for BIDW A Low Risk and Cost-Effective Approach to Architecting BIDW Environments Business Focused Develop & Deploy Extensive Help Ensure SLA Architecture Sound Attainment Knowledge, Proven BIDW Services Approach Architectures Delivery Methods • Executive Briefing • Workshop • Justification/TCO Analysis Workshop • Planning & Configuration • Training • Architecture & Design & POCs • Implementation & Uniquely satisfies Based on While meeting Help reduce Delivery BIDW specific business business cost & improve • Migration deployment requirements utilization requirements • Performance Tuning challenges
- 6. BI Platform – SAS & Sun
- 7. The SAS Industry Framework Covering End-To-End Business Intelligence Needs
- 8. SAS Enterprise BI Medical Services Home Office NTIS PPD Workstation Workstation IMS Reports Rep Laptop SFA OLTP Database Sales Analysis Customer Repository (Managed Data SAP SD Workstation Environment) Sales Admin HR Admin Workstation ZS Systems Workstation Cyborg SAP FI SAS 9 Accounting Workstation Cluster Analysis SAP HR ODS Concur Web Solutions ATSI Color- Newton CCT Blue Roche Reports graphics Current Dart MD Spin Dashboard Webserver Dart MC MCO Field IC Workstation E Initiatives Workstation Access1 BI SPIN
- 9. Sun BIDW Solution Architecture
- 10. Sun’ Infrastructure Options BIDW Industry Proof-of-Concept BIDW Services Leading Sun Service Sun Solution Architectures Partners Centers Availability Application Identity Communications Integration Web Suite Platform Management Suite Suite Infrastructure Suite Suite Suite Java Enterprise System Operating Systems Resource Data Management Management Processors/Servers Simplicity, Choice, Innovation, Security, Investment Protection
- 11. Data Integration
- 12. Data Integration 80 percent of BIDW projects are spent building and maintaining the data infrastructure
- 13. Requirements will drive the Design Define Business, System, Data, Process and Personnel CTQs ● What data and how often does the business require data to be loaded? ● What are the availability requirements? ● How much data will be pumped through the ETL process and what type of transformation will be done? ● How will this process be supported (application, operations, etc) and by whom?
- 14. Data Integration Design Alternatives Horizontal(Grid) vs Vertical: Operating Systems: Solaris(SPARC or x86), RedHat, Suse, Windows Chip Technologies: SPARC, CoolThreads, AMD, Intel Keys to Success: Focus on Requirements Focus on Supportability Focus on reducing complexity
- 15. Sun Strengths for Data Integration • CoolThreads – designed for throughput (8 cores, 32 threads) • UltraSPARC IV+ — world record setting performance(2.2TB < 90 minutes) Up to 32 Simultaneous • AMD Opteron x64 servers — 65 Threads! world records and counting • Sun Grid Computing – manage as a single resource, scalability • Manage individual resources with Solaris Containers
- 16. Data Warehouse
- 19. Once Again, Focus on Requirements Define Business, System, Data, Process and Personnel CTQs What are the business query requirements? (adhoc, analytics, basic reporting, etc) What are the availability requirements? (9x5, cost of downtime, etc.) What is the merge and purge requirements? (2 yrs online, 3 yrs within reach, etc.) How will this environment be supported (application, operations, etc) and by whom?
- 20. DW Design Alternatives/Opinions Clustered vs non-Clustered, Vertical vs Horizontal, Tiered Storage(ILM) vs non- Tiered Storage Focus on Data and System CTQs Limited Horizontally Scaling (AMD, SPARC) – (ie don’t scale past 4 nodes) Develop a ILM strategy up front to support your merge and purge strategy. Performance focused – remove dormant data
- 21. Sun Strengths for the Warehouse Enterprise Engine Ideal consolidation Designing next platform to reduce generation CPUs complexity (ROCK) for DB performance Proven performance for large DW—TPC-H QphH High Availability 59,435@3TB 31% > than On-the-fly leader scalability Servers and Storage
- 22. Service Availability for Every Need Availability
- 23. Sun Strengths for Data Lifecycle Management Multiple Tiers of Storage Partners
- 24. Storage Management • Automated storage resource management and provisioning • Data protection and recoverability • Hierarchical storage management – SAM-FS • Automated policy management – SAM-FS
- 25. Flexibility, Scalability, and Performance • Flexibility > Virtualization and automatic provisioning for data on-demand > Dynamically reconfigure resources > Solaris Containers and Solaris Resource Manager • Performance > World-record benchmarks for SMP systems > Sun is the fastest data warehouse engine > CMT – more throughput and scalability > MPO – faster access to memory Scalability • Scale up or out non-disruptively
- 26. Application Service Availability At a Low Cost – Sun Java Availability Suite • Continuous access to data and network • $50 USD/per/employee subscription model • Very fast recovery from failure • Support for Solaris SPARC and x64 platforms • Easier to manage Availability • Supports SAS products
- 27. Manageability • Dynamic domains - manage pools of resources individually • Solaris Containers - isolate applications in domains • Solaris Resource Manager - allocate resources where needed • Sun StorEdge Pool Manager - same concepts for storage • Sun StorEdge Data Services - common data protection services
- 28. BI Reporting & Analytics
- 29. BI Analysis Access to Multidimensional Data
- 30. BI Analysis Needs Access to Multidimensional Data SAS Oracle Cognos Hyperion Actuate Business Objects • Spontaneous, ad-hoc queries to uncover trends and identify anomalies • Support multi-threaded OLAP engines • N-tier for load balancing, high availability • 64-bit for larger analytical models
- 31. Considerations for BI Analysis • Flexibility and Scalability > UltraSPARC for vertical scaling to increase performance for analytics > CoolThreads for BI reporting > AMD Opteron servers for leading performance, low-cost, less space, and energy-efficient > J2EE architecture to support SOA initiatives > Managed as a single, unified resource
- 32. Common OS – Key Features The Solaris 10 • Binary compatibility and Operating scalability – 1 to 144 cpus System • Performance > Faster access to memory > Enhanced throughput > Streamlined networking stack • Virtualization and provisioning > Solaris Containers > Solaris Resource Manager > Dynamic System Domains, automated reconfiguration • Reliability > DTrace > Predictive Self Healing
- 33. BI Access Layer
- 35. Access and Delivery Needs • Secure access for customers, employees, and partners • Anywhere access • Central administration
- 36. Sun Strengths for Access and Delivery • Sun Java Identity Management Suite > Centralized enterprise wide approach > Single point of entry and sign-on > Quickly and automatically add and manage users > Control access to sensitive data > Secure, remote access • Customized Java Identity Management Suite for any BI tool • Sun & Teleran iSight and iGuard(Application Usage Mgmt) > Query monitoring, predictive analysis, and Web reporting > Query controls through policy management
- 37. How we can make a difference
- 38. Winter Top Ten Proven Performance • Sun has 6 of 10 largest DW on UNIX by normalized volume • Increased total awards from 28 to 72 in 2003 • 6 of 10, normalized data volume, UNIX, 5 of 10 all platforms • Top 6, including #1 OLTP highest workload – 8.6 million SQL statements per hour, top 3 Scientific • World's largest scientific system, normalized volume, 364 GB • World's largest commercial database, over 100 TB, runs on the Solaris OS
- 39. BIDW Proof-of-Concept Architectures Designed Reduces Risk Cost Pre-Tested Complexity
- 40. Sun Solution Centers Dedicated BIDW & SAS Competency Centers • World-wide • Proof-of-concept • Testing for performance and scalability • Partner offerings • Learn from experts
- 41. Summary • Focus your solution towards ALL requirements • Smarter solutions to help you stay competitive > Maximize investment in industry-leading technologies > Choice in partners for innovative, competitive solutions to solve your business needs > Reduce risk with a consistent end-to-end infrastructure, proof-of-concept architectures, and BIDW services > Open platform to integrate data and applications > Open, free, industrial-strength OS with the Solaris 10 OS > Industry-standard AMD Opteron servers running the Solaris OS, Linux, or Windows > One of the most comprehensive data, identity, and storage management portfolios in the industry
- 42. Next Steps • Visit sun.com/bidw • Contact your Sun Sales Representative to schedule a workshop • Visit a Sun Solution Center to test before you invest • Contact me at jim.mcmanus@sun.com
- 43. SUN BUSINESS INTELLIGENCE SOLUTIONS Jim McManus Managing Consultant - BI/DW Sun Microsystems, Inc.