Namik Hrle IBM Distinguished Engineer hrle@de.ibm.com Database Management System Trends IBM DB2 Perspective November, 2013 2013 IBM Corporation 2011 IBM Corporation
Disclaimer Copyright IBM Corporation 2013. All rights reserved. U.S. Government Users Restricted Rights - Use, duplication or disclosure restricted by GSA ADP Schedule Contract with IBM Corp. IBM s statements regarding its plans, directions, and intent are subject to change or withdrawal without notice at IBM s sole discretion. Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision. The information mentioned regarding potential future products is not a commitment, promise, or legal obligation to deliver any material, code or functionality. Information about potential future products may not be incorporated into any contract. The development, release, and timing of any future features or functionality described for our products remains at our sole discretion. IBM, the IBM logo, ibm.com, DB2, and DB2 for z/os are trademarks or registered trademarks of International Business Machines Corporation in the United States, other countries, or both. If these and other IBM trademarked terms are marked on their first occurrence in this information with a trademark symbol ( or ), these symbols indicate U.S. registered or common law trademarks owned by IBM at the time this information was published. Such trademarks may also be registered or common law trademarks in other countries. A current list of IBM trademarks is available on the Web at Copyright and trademark information at www.ibm.com/legal/copytrade.shtml Other company, product, or service names may be trademarks or service marks of others. 2
Agenda Business and Technology Drivers IBM DB2 Technology IBM DB2 Analytics Accelerator 3
Traditional Systems Landscape OLTP Staging Area ETL 4 ODS ETL EDW ETL Data Marts ETL
Traditional Systems Landscape OLTP Staging Area ETL 5 ODS ETL EDW ETL Data Marts ETL
Traditional Systems Landscape Applications operational OLTP Staging Area ETL 6 analytical ODS ETL EDW ETL Data Marts ETL
Traditional Systems Landscape Applications operational OLTP Staging Area ETL analytical ODS ETL EDW ETL Data Marts ETL Negative ramifications: Complexity both in systems management and in applications Difficulties in supporting real time analytics Inability to match ever more demanding SLA requirements High total cost of ownership 7
Traditional Systems Landscape Applications operational OLTP Staging Area ETL analytical ODS EDW ETL ETL Negative ramifications: Historical reasons: Complexity Different access patterns both in systems management and in applications Difficulties in supporting real time analytics Inability to match ever more demanding SLA requirements High total cost of ownership ETL impact on performance EDW as the data integration hub again, impact on performance Different life-cycle characteristics and again, impact on performance Different Service Level Agreements (SLA) 8 Data Marts Lack of broadly available workload management capabilities Choice of lower cost-of-acquisition offerings
Road to Visionary Systems Landscape Applications operational OLTP Staging Area ELT analytical ODS ELT EDW ELT Data Marts ELT Benefits Uniform policies and procedures for security, HA, DR, monitoring, same tools, same skills,... Efficient data movement within the system, often not involving network (ELT vs. ETL) Uniform access to any data for types of applications Opportunity to remove, i.e. consolidate some of the layers, ultimately leading to a single database 9
Visionary Systems Landscape Applications operational Benefits analytical Data Uniform policies and procedures for security, HA, DR, monitoring, same tools, same skills,... Efficient data movement within the system, often not involving network (ELT vs. ETL) Uniform access to any data for types of applications Opportunity to remove, i.e. consolidate some of the layers, ultimately leading to a single database 10
Visionary Systems Landscape Applications operational Benefits Uniform policies and procedures for security, HA, DR, monitoring, same tools, same skills,... analytical Data Challenges Efficient data movement within the system, often not involving network (ELT vs. ETL) Uniform access to any data for types of applications Opportunity to remove, i.e. consolidate some of the layers, ultimately leading to a single database 11 Mixed workload management capabilities Ensuring continuous availability, security and reliability Providing seamless scale-up and scale-out Providing universal processing capabilities to deliver best performance for both transactional and analytical workloads without the need for excessive tuning
Visionary Systems Landscape Applications operational Benefits Uniform policies and procedures for security, HA, DR, monitoring, same tools, same skills,... analytical Data Challenges Efficient data movement within the system, often not involving network (ELT vs. ETL) Uniform access to any data for types of applications Opportunity to remove, i.e. consolidate some of the layers, ultimately leading to a single database Mixed workload management capabilities Ensuring continuous availability, security and reliability Providing seamless scale-up and scale-out Providing universal processing capabilities to deliver best performance for both transactional and analytical workloads without the need for excessive tuning Approaches Large RAM 'In-memory' databases Massively parallel processing Large number of sockets, cores, servers Vector processing Hardware acceleration through special purpose processors FPGA, GPU,... Columnar stores Appliances 12
Re-inventing In-Memory Computing The benefits of in-memory processing have been known since the onset of IT itself The DIMM price per GB has decreased by 9.4 times since 2007 A genuine in-memory computing product must be designed with assumption that all (or most) of the data will be in memory at any point in time Most of the traditional database management systems do not satisfy this condition The key promises: 13 However, there are signs that the bottom of a down cycle might have been reached Supporting very large memory is not the same as supporting in-memory computing The fastest I/O is no I/O Many software components already support practically unlimited data cache sizes The limiting factor has been the cost lightning performance without tuning eliminating all the data redundancy that is traditionally created to deliver acceptable performance
In-Memory Database Challenges Database Management System is a state-full resource and non-volatile storage is realistically unavoidable for fast and reliable recovery Many wrongly conclude that an in-memory database does not require any disk storage 14 particularly challenging for enterprise data warehouses incompatible with Big Data requirements Scale out is typically based on shared-nothing architecture For example, SAP's HANA requires much more disk storage than the real memory 'Data must fit in memory' is a major limitation If there is no disk, the database would not be recoverable Writing logs to disk happens at commit Writing data to disk happens periodically (checkpoints, savepoints,...) Does not scale well for workloads that do not adhere to data-to-node affinity DRAM is still much more expensive than disk New comers will need to address many non-performance quality of service characteristics Eliminating data redundancy is not a realistic goal due to integration hub aspects of enterprise data warehouses
Row-oriented vs. Column-oriented Data Store Model SELECT * FROM T WHERE C1 = unique key R1 C1 C2 C3 C4 SELECT AVG(C3) FROM T C5 C1 C2 C3 C4 C3 C4 C5 R2 R3 R4 Row Store C1 C2 C3 C4 C1 C5 R1 RR11 R2 RR22 R3 RR33 R4 RR44 C2 C5 Column Store 15
Agenda Business and Technology Drivers IBM DB2 Technology IBM DB2 Analytics Accelerator 16 IBM Internal Use
Visionary Systems Landscape Applications operational Benefits Uniform policies and procedures for security, HA, DR, monitoring, same tools, same skills,... analytical Data Challenges Efficient data movement within the system, often not involving network (ELT vs. ETL) Uniform access to any data for types of applications Opportunity to remove, i.e. consolidate some of the layers, ultimately leading to a single database Mixed workload management capabilities Ensuring continuous availability, security and reliability Providing seamless scale-up and scale-out Providing universal processing capabilities to deliver best performance for both transactional and analytical workloads without the need for excessive tuning Approaches Large RAM 'In-memory' databases Massively parallel processing Large number of sockets, cores, servers Vector processing Hardware acceleration through special purpose processors FPGA, GPU,... Columnar stores Appliances 17
Visionary Systems Landscape Applications operational Benefits Uniform policies and procedures for security, HA, DR, monitoring, same tools, same skills,... Data Challenges Efficient data movement within the system, often not involving network (ELT vs. ETL) Uniform access to any data for types of applications Opportunity to remove, i.e. consolidate some of the layers, ultimately leading to a single database Approaches analytical Large RAM 'In-memory' databases Massively parallel processing Large number of sockets, cores, servers Vector processing Hardware acceleration through special purpose processors FPGA, GPU,... Columnar stores Appliances 18 Mixed workload management capabilities Ensuring continuous availability, security and reliability Providing seamless scale-up and scale-out Providing universal processing capabilities to deliver best performance for both transactional and analytical workloads without the need for excessive tuning Building on proven technology base DB2 (both z/os and LUW) already provide superior technology to address most of the challenges The remaining challenge is addressed by adding special purpose processing component for analytical workloads DB2 for z/os: IBM DB2 Analytics Accelerator DB2 for LUW: BLU
DB2 for z/os Approach: Hybrid Database Management System Applications DBA Tools, z/os Console,... Application Interfaces (standard SQL dialects) Operation Interfaces (e.g. DB2 Commands) DB2 Data Manager 19 Buffer Manager... IRLM Log Manager IBM DB2 Analytics Accelerator
DB2 for LUW Approach: BLU Acceleration New innovative technology for analytic queries Dynamic in-memory technology loads terabytes of data in RAM instead of hard disks. This streamlines query workloads even when data sets exceed the size of the memory. Columnar store scans and locates the most relevant data based on columns instead of rows, resulting in faster processing. New run-time engine exploits cache-aware memory management and parallel vector processing providing multi-core and multiple data parallelism (SIMD) and allowing you to analyze data in parallel over different processor sockets and cores Actionable compression enables data to be analyzed in compressed format and results in further storage reduction Data skipping skips unnecessary processing of irrelevant or duplicate data, loading only the information that needs to be analyzed. Revolution by Evolution Built directly into the DB2 kernel BLU tables can coexists with traditional row tables, in same schema, tablespaces, bufferpools Query any combination of BLU or row data Memory-optimized (not in-memory ) Value : Order-of-magnitude benefits in Performance Storage savings Time to value 20
DB2 Family Different code bases DB2 for z/os is written in PL/X and runs on z/os DB2 for LUW is written in C and runs on multiple operating systems but very close cooperation between development teams Common application interfaces Common SQL interface DB2 SQL Language Council ensures consistency is maintained DB2 SQL Reference for Cross Platform Development (over 1200 pages) Starburst optimizer purexml System z Parallel Sysplex and LUW purescale Bitemporal data Row and column access control On-going work on uniform database administration tasks Tools like Optim Data Studio 21
Agenda Business and Technology Drivers IBM DB2 Technology IBM DB2 Analytics Accelerator 22 Built on DB2 - The Industrial Strength DBMS Architecture Customer References Powered by PureData for Analytics
Synergy with System z Capacity on Demand and backup and recovery solutions lets you be more responsive to your needs, frees your staff up to do more important work "Shared data" database environment and synergies with z/os means data is more available Robust z/os allows database serving without interruption, even in the event of an operating system function error System z Philosophy: The more errors prevented at the hardware and microcode levels the less impact on applications, operations, and end users Highest availability on the planet Continuous availability Non-disruptive upgrades of hardware, operating system, applications and database systems Comprehensive multi-site disaster recovery Unmatched end-to-end security from logon through data encryption System-level mixed workload management with full resource utilization Special component named the Workload Manager manages all resources 100% utilization, 24 hours a day Most cost effective SLA The most cost effective platform to manage and maintain 23
DB2 and zenterprise EC12 Faster CPU 1.25x compared to z196 20-28% CPU reduction measured with DB2 OLTP workloads 25% reduction measured with DB2 query and utilities workloads Less compression overhead with DB2 data (1-15%) 50% More System Capacity to help consolidation Excellent synergy with DB2 10 scalability New Features DB2 11 plans to exploit FLASH memory and pageable 1MB frames 2GB frame support drive additional CPU savings DB2 code backed by large frames for CPU reductions Enhanced prefetch instruction for CPU reductions Transactional Memory provides further possibilities for performance gains 24
DB2 11 Major Themes Performance Improvements Continuous Availability Features Faster, more efficient performance for query workloads Temporal and SQLPL enhancements Transparent archiving SQL improvements and IDAA enhancements Simpler, faster DB2 version upgrades 25 Improved autonomics which reduces costs and improves availability Making online changes without affecting applications Online REORG improvements, less disruption DROP COLUMN, online change of partition limit keys Extended log record addressing capacity (1 yottabyte) BIND/REBIND, DDL break into persistent threads Enhanced business analytics Improving efficiency, reducing costs, no application changes 0-5% for OLTP, 5-15% for update intensive batch 5-20% for query workloads Less overhead for data de-compression Exploitation of new zec12 hardware features No application changes required for DB2 upgrade Access path stability improvements Product stability: support pre GA customer production
DB2 for z/os and Distributed BigData DB2 is providing the connectors and the DB capability to allow DB2 applications to easily and efficiently access data in Hadoop HDFS_Read New user-defined functions New generic table UDF capability JAQL_Submit 26 HDFS_Read is a user-defined table function to read a file in Hadoop file system. The output schema is determined at query time. JAQL_Submit is a user-defined scalar function to submit a JAQL script to BigInsight
Agenda Business and Technology Drivers IBM DB2 Technology IBM DB2 Analytics Accelerator 27 Built on DB2 - The Industrial Strength DBMS Architecture Customer References Powered by PureData for Analytics
DB2 Components Applications DBA Tools, z/os Console,... Application Interfaces (standard SQL dialects) Operation Interfaces (e.g. DB2 Commands) DB2 Data Manager 28 Buffer Manager... IRLM Log Manager
IBM DB2 Analytics Accelerator as a Virtual DB2 Component Applications DBA Tools, z/os Console,... Application Interfaces (standard SQL dialects) Operation Interfaces (e.g. DB2 Commands) DB2 Data Manager 29 Buffer Manager... IRLM Log Manager Accelerator
DB2 Becomes a Hybrid Database Management System Applications DBA Tools, z/os Console,... Application Interfaces (standard SQL dialects) Operation Interfaces (e.g. DB2 Commands) DB2 Data Manager 30 Buffer Manager... IRLM Log Manager IBM DB2 Analytics Accelerator
Connectivity Options Multiple DB2 systems can connect to a single accelerator A single DB2 system can connect to multiple accelerator DB2 Accelerator Accelerator DB2 DB2 Accelerator Accelerator Multiple DB2 systems can connect to multiple accelerator DB2 DB2 Accelerator Policy based workload management Better utilization of accelerator resources Scalability High availability 31 Full fl exibility for DB2 systems: residing in the same LPAR residing in different LPARs residing in different CECs being independent (non-data sharing) belonging to the same data sharing group belonging to different data sharing groups
DB2 for z/os: Query Execution Process Flow Application Interface Heartbeat Optimizer SPU CPU FPGA Query execution run-time for queries that cannot be or should not be off-loaded to IDAA SPU CPU SMP Host Application IDAA DRDA Requestor Memory FPGA Memory SPU CPU FPGA Memory SPU CPU FPGA Memory DB2 for z/os IDAA Queries executed without IDAA Queries executed with IDAA Heartbeat (IDAA availability and performance indicators) 32
Data Synchronization Options Synchronization options Use cases, characteristics and requirements Full table refresh Existing ETL process replaces entire table The entire content of a database table is refreshed for accelerator processing Multiple sources or complex transformations Smaller, un-partitioned tables Reporting based on consistent snapshot Need for refresh automatically detected Table partition refresh For a partitioned database table, selected partitions can be refreshed for accelerator processing Optimization for partitioned warehouse tables, typically appending changes at the end More efficient than full table refresh for larger tables Reporting based on consistent snapshot Need for refresh automatically detected Incremental update Scattered updates after bulk load Log-based capturing of changes and propagation to IDAA with low latency (typically few minutes) Reporting on continuously updated data (e.g., an ODS), considering most recent changes More efficient for smaller updates than full table refresh Applications can request reporting on committed data only 33
High Availability Configuration IDAA 1 Tab 1 Tab 2 Tab 3 Tab 1 Tab 2 Tab 3 Tab 5 Tab 4 System z DB2 for z/os IDAA 2 Tab 1 Tab 2 Tab 3 34
Disaster Recovery Configuration Example: Prior to Disaster Site A Tab 1 Tab 5 Tab 3 Tab 4 Tab 2 Site B synchronous replication Tab 1 Tab 5 Tab 3 Tab 4 Tab 2 System z System z DB2 DB2 CF Member 1 CF Member 2 IDAA 1 Tab 1 IDAA 2 Tab 1 Tab 2 Tab 2 Tab 3 35 Tab 3
Disaster Recovery Configuration Example: Disaster Happens Site A Tab 1 Tab 5 Tab 3 Tab 4 Tab 2 Site B synchronous replication Tab 1 Tab 5 Tab 3 Tab 4 Tab 2 System z System z DB2 DB2 CF Member 1 CF Member 2 IDAA 1 Tab 1 IDAA 2 Tab 1 Tab 2 Tab 2 Tab 3 36 Tab 3
Disaster Recovery Configuration Example: After Disaster Site A Tab 1 Tab 5 Tab 3 Tab 4 Tab 2 Site B Tab 1 Tab 5 Tab 3 Tab 4 Tab 2 System z System z DB2 DB2 CF Member 1 CF Member 2 IDAA 1 Tab 1 IDAA 2 Tab 1 Tab 2 Tab 2 Tab 3 37 Tab 3
High Performance Storage Saver Major saving of host disk space for historical data Historical Data Year Year -1 Year -2 Year -3 Year -4 Year -5 Year -7 1Q 1Q 1Q 1Q 1Q 1Q 1Q 2Q 2Q 2Q 2Q 2Q 2Q 2Q 3Q 3Q 3Q 3Q 3Q 3Q 3Q 4Q 4Q 4Q 4Q 4Q 4Q Current Data 4Q One Quarter = 3.57% of 7 years of data One Month = 1.12% of 7 years of data One month = 2.78% of 3 years of data 38
High Performance Storage Saver Storing historical data in accelerator only DB2 Query from Application Part #1 No longer present on DB2 Storage Or Part #1 Part #2 Part #3 Active Accelerator Part #4 Part #5 Part #6 Part #7 Historical Time-partitioned tables where: only the recent partitions are used in a transactional context (frequent data changes, short running queries) the entire table is used for analytics (data intensive, complex queries). High Performance Storage Saver s Archive Process: 39 Data is loaded into Accelerator if not already loaded Automatically takes Image Copy of each partition to be archived Automatically remove data from DB2 archived tablespace partitions DBA starts archived partitions as read-only
Data Residency to Match Query Types Applications SQL DB2 Table A DB2 DB2 Table A Table A Accelerator Accelerator Table A Active Table A Active & Historical Query Types Transactional only Mixed workload Active data only Active data only Historical data only Active & historical data Mixed workload 40 20 1 3 IBM Corporation 20 1 1 IBM Corporation
Agenda Business and Technology Drivers IBM DB2 Technology IBM DB2 Analytics Accelerator 41 Built on DB2 - The Industrial Strength DBMS Architecture Customer References Powered by PureData for Analytics
Customer References Major US Healthcare Insurance Company The company provides a range of insurance products and related services for 10s of millions of members. The network includes 100,000s of doctors, 1000s of hospitals and nearly a million other healthcare professionals. Business challenge The changing healthcare landscape drove the company to ensure it could manage a massive influx of data and the mounting reporting requirements as the Affordable Care Act ushers tens of millions of new customers into the insurance market. Customer Benefits Enables the company to meet stringent on-time reporting requirements with the solution s incremental update feature The DB2 Analytics Accelerator greatly exceeded our expectations. The first time we ran our very resource-intensive queries on the solution, queries which had historically taken hours to run, they ran in seconds. - Systems Engineering Manager Anticipates a significant reduction in storage costs with the data server s high-performance storage saver feature Processes some queries thousands of times faster, reducing query times from nearly 3 hours to 6 seconds 42
Customer References Example: 300 Mixed Workload Queries 270 queries continue to execute in DB2 returning results in seconds or subseconds 30 complex, expensive queries got routed to IDAA and reduced elapsed time and CPU cost by orders of magnitude. Query Query 1 Query 2 Query 3 Query 4 Query 5 Query 6 Query 7 Query 8 Query 9 43 Customer Table ~ 5 Billion Rows Total Total Rows Rows Reviewed Returned 2,813,571 853,320 2,813,571 585,780 8,260,214 274 2,813,571 601,197 3,422,765 508 4,290,648 165 361,521 58,236 3,425.29 724 4,130,107 137 Times Faster DB2 Only DB2 with IDAA Hours Sec(s) 2:39 9,540 2:16 8,220 1:16 4,560 1:08 4,080 0:57 4,080 0:53 3,180 0:51 3,120 0:44 2,640 0:42 2,520 Hours Sec(s) 5 0.0 5 0.0 6 0.0 5 0.0 70 0.0 6 0.0 4 0.0 2 0.0 193 0.1 1,908 1,644 760 816 58 530 780 1,320 13
Customer References Large European Insurance Company Business challenge: With roughly 2.5 billion transactions in the company s financial data store, fast and accurate analysis is essential for setting the right premiums. To improve access to claims data across its multiple international locations, the company needs to increase system availability, optimize workloads, speed queries and accelerate the generation of claims reports run by internal business users. Solution: Deploy the IBM zenterprise System with DB2 for z/os to process all data loads from a central location, and IBM DB2 Analytics Accelerator to deliver faster responses to individual analytic queries. We were surprised by the performance gain IBM DB2 Analytics Accelerator provided, as well as its ability to further boost the capacity of our IBM zenterprise System. - Director of Operations Benefits Speeds report generation by as much as 70 percent through faster query response time, and improves staff efficiency by centralizing data on a single platform Reduces processing costs and CPU consumption by routing eligible workloads to the accelerator Increases satisfaction among internal business users by delivering a comprehensive overview of claims transactions that integrates operational data with advanced analytics 44
Customer References Major Italian Bank Business challenge: One of the largest banks in Italy. Employing 1,000s of people and generating annual revenue of 100s of millions, the group provides banking, insurance and asset management services from more than 1000 branches across the country. They wanted to meet their growth objectives by identifying customer demand for new products or services, then adapting their offerings to win the new business. The challenge was extracting actionable insight from its big data, as the size of its databases made queries from business users frustratingly slow. Solution: The bank created the big data project an initiative to develop the infrastructure to support the analytics requirements of the business. As a first step, the bank implemented IBM DB2 Analytics Accelerator on its existing IBM System z mainframes. The DB2 Analytics Accelerator inherits all of the benefits of System z including security, performance and scalability Benefits Offers rapid time-to-insight for 1,000 business users informing the development of new products, services and strategies. Enables the bank to match its offering with customer demand driving business growth in line with corporate objectives. Creates a platform for future innovation, including data mining from IBM SPSS and marketing management from IBM Campaign. 45 Being a leader in the banking industry requires a strong commercial offering that meets fast-evolving customer expectations. To understand what your customers want, you need an excellent grasp of your business data, and to develop new products and services, you need the ability to deliver those insights rapidly to the right people in the business. - Chief Information Officer Video Link
Customer References Large Central European Bank SQL DB2 on z196 Standalone Netezza Exadata DB2 with DB2 Analytics Accelerator Query 1 00:01:50 00:00:04 00:00:09 00:00:03 Query 2 00:75:31 00:00:09 00:00:39 00:00:04 Query 3 00:00:46 00:00:05 00:00:13 00:00:02 Business challenge: Experienced performance issues with its data warehouse. Required to supply financial activity reports to European Central Bank (ECB) by 9 am every business day. Performance issues were seriously hindering bank s ability to meet this objective. The bank needed a technology solution that would address and eliminate performance issues and enable timely financial reporting to support compliance requirements. Benefits: Less time for tuning of SQL statements No data base maintenance define tables/refresh data Faster, more agile development Coexistence of OLTP and DWH databases on same LPAR CPU saving because of redirecting execution to IBM DB2 Analytics Accelerator 46 The IBM DB2 for z/os is a secure and highly available repository for the bank's data. Highperformance specialty processors have significantly improved query response times as compared to our previous solution. The new zenterprise hybrid technology is highly scalable and flexible which means that our users are now able to access the information they need more quickly. Chief Information Officer
Agenda Business and Technology Drivers IBM DB2 Technology IBM DB2 Analytics Accelerator 47 Built on DB2 - The Industrial Strength DBMS IBM DB2 Analytics Accelerator Architecture Customer References Powered by PureData for Analytics
Powered by PureData System for Analytics N2001 12 Disk Enclosures 288 600 GB SAS2 Drives 240 User Data, 14 S-Blade 34 Spare RAID 1 Mirroring 2 Hosts (Active-Passive) 2 6-Core Intel 3.46 GHz CPUs 7x300 GB SAS Drives Red Hat Linux 6 64-bit Scales from ½ Rack to 4 Racks 7 PureData for Analytics S-Blades User Data Capacity: Data Scan Speed: 2 Intel 8 Core 2+ GHz CPUs 2 8-Engine Xilinx Virtex-6 FPGAs 128 GB RAM + 8 GB slice buffer Linux 64-bit Kernel 192 TB* 478 TB/hr* Power Requirements: Cooling Requirements: * 4X compression assumed 48 7.5 kw 27,000 BTU/hr
N2001 Snippet-BladeTM (S-Blade) Components IBM BladeCenter Server HX5 Blade 128 GB RAM 16 Intel cores Netezza DB Accelerator 49 BPE4 Side Car 16 GB RAM 16 Virtex-6 FPGA cores SAS Controller
N2001: Speed Through Taking Most of Streaming Capabilities 2.5 drives per core 325 MB/s DB2 for z/os CPU Core FPGA Core 1000 MB/s 1000 MB/s 130 MB/s 1300 MB/s 130 MB/s 4x compression assumed Complex Joins, Aggs, etc. 50 Restrict Visibility Project Decompress S-Blade Table Cache 65 MB/s
IBM PureData System for Analytics Models Comparison N1001 N2001 Blade type HS22 HX-5 CPU sockets & cores per blade 2 x 4 Core Intel CPUs 2 x 8 Core Intel CPUs # Disks 96 x 3.5 / 1 TB SAS (92 Active) 288 x 2.5 / 600GB SAS2 (240 Active) Raw Capacity 96 TB 172.8 TB Total Disk Bandwidth ~11 GB/s ~32 GB/s S-Blades per Rack (cores) 14 (112) 7 (112) S-Blade Memory 24 GB 128 GB Rack Configurations ¼, ½, 1, 1 ½, 2, 3, 10 ½, 1, 2, 4 FPGA Cores / Blade 8 (2 x 4 Engine Xilinx FPGA) 16 (2 x 8 Engine Xilinx Virtex 6 FPGA) 128 TB 192 TB User Data / Rack (assuming 4x compression) 51
IBM DB2 Analytics Accelerator Supports All Models N2001 Models 005 010 025 040 Cabinets 1/2 1 2 4 S-Blades 4 7 14 28 Processing Units 64 112 224 448 Capacity (TB) 24 48 96 192 Effective Capacity (TB)* 96 192 384 768 N1001 Models 002 005 010 015 025 030 040 060 080 100 Cabinets ¼ ½ 1 1½ 2 3 4 6 8 10 S-Blades 3 6 12 18 24 36 48 72 96 120 Processing Units 24 48 96 144 192 288 384 576 768 960 Capacity (TB) 8 16 32 48 64 96 128 192 256 320 Effective Capacity (TB)* 32 64 128 192 256 384 512 768 1024 1280 Capacity = User data space Effective Capacity = User data space with compression (4x compression assumed) 52
Agenda Business and Technology Drivers IBM DB2 Technology IBM DB2 Analytics Accelerator 53 Built on DB2 - The Industrial Strength DBMS Architecture Customer References Powered by PureData for Analytics Strategy and Roadmap
Strategy Enable DB2 transition into a truly universal DBMS that provides best characteristics for both OLTP and analytical workloads. DB2 for z/os ELT Accelerator Advanced Analytics Query Accelerator Storage Saver OLTP in-database transformations advanced analytical capabilities multi-temperature and storage saving solutions 54 Complement DB2's industry leading transactional processing capabilities Provide specialized access path for data intensive queries Enable real and near-real time analytics processing Execute transparently to the applications Operate as an integral part of DB2 and System z Reusing industry leading PDA's query and analytics capabilities and take advantage of future enhancements Extend query acceleration to new, innovative usage cases, such as: Ultimately allow consolidation and unification of transactional and analytical data stores
Roadmap Unified Store supporting new use cases increasing IDAA transparency enabling more query acceleration enhancing current capabilities p Im e m e r ov of s nt fe g tin s i ex s e r atu Advanced Analytics ELT Accelerator V4 Storage Saver V2 A PD h c e t lo no g vo e y on i t lu V3 Query Accelerator V1 55
Fast Evolution of IBM DB2 Analytics Accelerator Version 1 Nov 2010 IBM Smart Analytics Optimizer In-memory, column-store, multi-core and SIMD algorithms Discontinued and replaced by IBM DB2 Analytics Accelerator Version 2 Nov 2011 New name: IBM DB2 Analytics Accelerator Incorporates Netezza query engine Preserves key V1 value propositions and adds many more Version 3 Nov 2012 Better performance, more capacity Incremental update High Performance Storage Server Version 4 56 Nov 2013 Much broader acceleration opportunities More enterprise features
IDAA V3 Highlights Generally available since November 2012 Propagating DB2 changes to the accelerator as they happen: Incremental Update Reducing disk storage cost by archiving data in the accelerator and maintaining the excellent performance for analytical queries: High Performance Storage Saver Workload Manager integration Automatic detection of needs to refresh data in the accelerator More query routing control for applications (all, eligible) More query offload (e.g. DB2 OLAP functions) Speeding-up data refresh and reducing associated CPU cost on System z Accelerating in-database transformation (1) Enhancing high availability and scaling out (1) Improving performance of queries that generate very large result sets Supporting multi-byte EBCDIC data encoding (phase 1) (1) Increasing capacity to more than 1 petabyte Support for SAP workloads (1) (1) (1) (1) (1) features retrofitted to V2 57
IDAA V3 Highlights Additions since GA Additional query engine: PureData System for Analytics N2001 Support for Netezza operating system 7 Further reduction of CPU time associated with IDAA load process Up to 30% Enhancements in DFSMS BSAM routines managing data on the USS pipes z/os PTFs: z/os V1.12 UA68971 z/os V1.13 UA68972 z/os V2.1 UA68973 Multiple time zones in the same accelerator Limited support for LOCAL DATE setting Support for BITAND and TIMESTAMPDIFF functions Support for DECFLOAT when used as implicit cast e.g. when comparing different data types 58 Enhancements to incremental update
Version 4 at a Glance More Query Acceleration Static SQL DB2 11 (2) Multi-row fetch from local applications EBCDIC and Unicode in the same DB2 system and accelerator Enhanced Capabilities Improved Transparency Greatly improved scalability of Incremental Update Better performance of Incremental Update Improved performance for large result sets (2) Better access control for HPSS archived partitions Automatic workload balancing with multiple accelerators New RTS 'last-changed-at' timestamp (2) Automated NZKit installation Built-in Restore for HPSS HPSS archiving to multiple accelerators Protection for image copies created by HPSS archiving process Extending WLM support to local applications Profile controlled special registers (2) Rich system scope monitoring Improved continuous operations for Incremental Update Reporting prospective CPU cost and elapsed time savings Separation of duties for accelerator system administration operations Loading from flat file or image copy (1) Loading in parallel to DB2 and accelerator (1) Loading data as of any past point in time (1) Loading data to accelerator only (1) Enabling 59 new use (1) delivered by a separate tool (2) DB2 11 only cases