Intro to Query Optimization DB2 UDB for iSeries

1. Intro to Query OptimizationDB2 UDB for iSeries Tom McKinley IBM Rochester, MN USA

2. Background / Foundation

3. IBM's DB2 UDB Family • Three code bases... • Based on the system history, architecture and operating system • DB2 UDB for Linux, UNIX, Windows (LUW) • DB2 UDB for z/OS (S/390) • DB2 UDB for iSeries (AS/400)

4. DB2 UDB for iSeries • i5 + i5/OS • System viewed as a database server, not just an application system • DB2 UDB for iSeries (integrated part of OS/400 or i5/OS) • Universal Database support • Data Centric focus • Business logic moving into the database engine • SQL (DDL and DML) as primary interface to database • GUI to operating system and database via iSeries Navigator

5. iSeries - Logical Partitioning (LPAR) LPAR-1 LPAR-2 LPAR-3 IXS/IXA i5/OS Linux AIX Windows*** DB2 UDB for iSeries DB2 UDB for Linux DB2 UDB for AIX DB2 UDB for Win Virtual 1Gbit Ethernet LAN *** No LPAR support

6. iSeries i5 i5/OS Architecture M E M O R Y Multiple CPUs N-way SMP QUERY Single System Single Level Storage 64 bit POWER Storage Management IOP IOP IOP IOP IOP IOP IOP Table

7. i5/OS Objects SQL i5/OS library schema/collection physical file table logical file view keyed logical file index record row field column journal log

8. i5/OS Objects SELECT... FROM Physical File Library (Schema) CREATE ALIAS... Physical File (Table) Member 1 SELECT... FROM Alias_1 Alias_1 Member 2 SELECT... FROM Alias_2 Alias_2 Member 3 SELECT... FROM Alias_3 Alias_3

9. i5/OS Objects • System • Library • Object • Type • Attribute (subtype) • System • My_Schema • DB_Table • *FILE • PF (physical file) • System • My_Schema • DB_Index • *FILE • LF (logical file) • System • My_Schema • DB_View • *FILE • LF (logical file) Must be unique CREATE TABLE My_Schema.DB_Table ... CREATE INDEX My_Schema.DB_Index ... CREATE VIEW My_Schema.DB_View ...

10. i5/OS Objects One Database Management System with multiple interfaces • Structured Query Language (SQL) • Embedded • ODBC • JDBC • CLI Command Language (CL) DB2 DB File (PF) object CREATE TABLE CRTPF SELECT... FROM... High Level Language Native I/O

11. SQL Query Processing SQL request Optimize Open Run DB2 UDB for iSeries

12. Query Optimization

13. V5R1 Database Architecture ODBC / JDBC / ADO / DRDA / XDA Network Host Server CLI / JDBC Static Dynamic Extended Dynamic The optimizer and database engine are separated at different layers of the operating system Compiled embedded statements Prepare once and then reference Prepare every time SQL Native (Record I/O) Optimizer DB2 UDB (Data Storage & Management)

14. V5R2 and V5R3 Database Architecture ODBC / JDBC / ADO .NET / DRDA / XDA Network Host Server CLI / JDBC The optimizer and database engine merged to form the SQL Query Engine, and much of the work was moved to SLIC Static Dynamic Extended Dynamic Compiled embedded statements Prepare once and then reference Prepare every time SQL Native (Record I/O) Optimizer DB2 UDB (Data Storage & Management)

15. V5R2 and V5R3 Database Architecture

16. The Query Dispatcher • Determines which engine will optimize and process each query request • Only SQL requests are considered for the SQL Query Engine • Initial step for all query optimization that occurs in i5/OS • Ability to “back up” and use the Classic Query Engine when non-standard indexes are encountered during optimization • Initial goal is to use SQE

17. The Query Dispatcher – V5R2 • Dispatched to CQE if: • >1 Table (i.e. no joins) • OR & IN predicates • SMP requested • Non-Read (INSERT with subselect can use new path) • LIKE predicates • UNIONS • View or Logical File references • Subquery • Derived Tables & Common Table expressions, UDTFs • LOB columns • LOWER, TRANSLATE, or UPPER scalar function • CHARACTER_LENGTH, POSITION, or SUBSTRING scalar function using UTF-8/16 • Sort Sequences & CCSID translation between columns • Distributed queries via DB2 Multisystem • Non-SQL queries (QQQQry API, Query/400, OPNQRYF) • ALWCPYDTA(*NO) specified • Sensitive Cursor SQE support added into V5R2 - May 2003 (Latest DB Group + SI07650) Not part of any package

18. The Query Dispatcher - V5R3 • Dispatched to CQE if: • LIKE predicates • Logical File references • UDTFs • LOB columns • LOWER, TRANSLATE, or UPPER scalar function • CHARACTER_LENGTH, POSITION, or SUBSTRING scalar function using UTF-8/16 • Sort Sequences & CCSID translation between columns • DB2 Multisystem • Non-SQL queries (QQQQry API, Query/400, OPNQRYF) • ALWCPYDTA(*NO) specified • Sensitive Cursor • SQE now optimizes • VIEWS, UNIONS, SubQueries • INSERT, UPDATE, DELETE • Star Schema Join queries • Only SQE optimizes • INTERSECT • EXCEPT

19. The Query Dispatcher • Back up to CQE to complete optimization if any of the following are encountered: • Select/omit logical file • Logical file over multiple members • Join logical file • Derived key (s) • Native logical files that perform some intermediate mapping of the fields referenced in the key. Common ones are renaming fields, adding a translate or only selecting a subset of the columns • Specifying an Alternate Collating Sequence (ACS) on a field used for a key will also make a “derived key” (an implied map occurs within the index) • Sort Sequence (NLSS) specified for index or logical file • Probably the trickiest one to detect for users. The index is built while an NLSS table is specified in the query environment • Cost to “back up” and revert to CQE adds about 15% to the total optimization time • QAQQINI parameter to ignore unsupported logical files • Ignore_Derived_Index = *YES

20. Optimization • The Optimizer • Writes the best? program to fulfill your request • The Optimizer • Provides the recipe • Provides the methods • Does no cooking

21. Optimization... the intersection of various factors Server attributes Server configuration Version/Release/Modification Level Server performance The Plan SMP Database design Job, Query attributes Table sizes, number of rows SQL Request Static Dynamic Extended Dynamic Interfaces Views and Indexes (Radix, EVI) Work management

22. (Query) Access Plans • The output of query optimization (“the recipe and methods”) • Contents • A control structure that contains information on the actions necessary to satisfy each SQL request • These contents include: • Access Method • Info on associated tables and indexes • Any applicable program and/or environment information

23. Query Optimization • Cost Based Query Optimization • The DB2 for iSeries Optimizer performs "cost based" optimization • "Cost" is defined as the estimated time it takes to run the request • "Costing" various plans refers to the comparison of a given set of algorithms and methods in an attempt to identify the "fastest" plan • Optimization is based on time, not on resource utilization • Usually the fastest plan is also the most resource efficient plan, but this is not necessarily true • The goal of the optimizer is to eliminate I/O as early as possible by identifying the best path to and through the data • The optimizer has the ability and freedom to "rewrite" the query

24. Query Phases • Query processing can be divided into four phases: • Query Validation • Validate the query request • Validate existing access plan • Builds internal query structures • Query Dispatcher • Determine which query engine should complete the processing • Query Optimization • Choose most efficient access method • Builds access plan • Query Execution • Build the structures needed for query cursor • Build the structures for any temporary indexes (if needed) • Builds and activates query cursor (ODP) • Generate any feedback requested • Debug messages in the job log • DB Monitor records • Visual Explain We can affect this...

25. Query Optimization Feedback SQE Plan Cache Visual Explain DB Monitor Data SQL request Joblog Messages Query Optimization SQL Info from PGMs & PKGs

26. Data Access Methods Cost based optimization dictates that the fastest access method for a given table will vary based upon selectivity of the query High Response Time Method 3 Method 2 Method 1 Low Few Many Number of rows searched / accessed

27. Strategy for Query Optimization • Query optimization will generally follow this simplified strategy: • Gather meta-data and statistics for costing • Selectivity statistics • Indexes available to be costed • Sort the indexes based upon their usefulness • Environmental attributes that may affect the costs • Generate default cost • Build an access plan associated with the default plan • For each index: • Gather information needed specific to this index • Build an access plan based on this index • Cost the use of the index with this access plan • Compare the resulting cost against the cost from the current best plan

28. Strategy for Query Optimization • Optimizing indexes will generally follow this simplified strategy: • Gather list of indexes for statistics and costing • Sort the list of indexes considering how the index can be used • Local selection • Joining • Grouping • Ordering • Index only access • One index may be useful for statistics, and another useful for implementation

29. Statistics • All query optimizers rely upon statistics to make plan decisions • DB2 UDB for the iSeries has always relied upon indexes as its source for stats • Other databases rely upon manual stats collection for their source • SQE offers a hybrid approach where column stats will be automatically collected for cases where indexes do not already exist

30. Sources of Information • Meta-data sources • Existing indexes (Radix or Encoded Vector) • More accurately describes multi-column key values • Stats available immediately as the index maintenance occurs • Selectivity estimates from radix by reading n keys • Selectivity from EVI by reading symbol table values • Column Statistics • SQE only • Column Cardinality, Histograms & Frequent Values List • Constructed over a single column in a table • Stored internally as a part of the table object after created • Collected automatically by default for the system • Stats not immediately maintained as the table changes • Stats are refreshed as they become “stale” over time • Default sources • No representation of actual values in columns Best Worst

31. SQE Automatic Stats Collection • i5/OS Statistics collection job • Reactive, based on query requests • Automatic collection runs in this background job at very low priority • QDBFSTCCOL system job • Statistics Manager continuously analyzes entries in the Plan Cache and queues up requests for the collection job • Controlled by system value QDBFSTCCOL • iSeries Navigator graphical interface to manage stats collected by the system • API’s also provided to manage the stats

32. Review • What is the optimizer's job? • What is the optimizer's output? • What are some of the key elements used for cost based optimization? • What things affect the Access plan? • Look at resources used as well as response time.

33. Trademarks and Disclaimers IBM Corporation 1994-2005. All rights reserved. References in this document to IBM products or services do not imply that IBM intends to make them available in every country. The following terms are trademarks of International Business Machines Corporation in the United States, other countries, or both: Rational is a trademark of International Business Machines Corporation and Rational Software Corporation in the United States, other countries, or both. Java and all Java-based trademarks are trademarks of Sun Microsystems, Inc. in the United States, other countries, or both. Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both. Intel, Intel Inside (logos), MMX and Pentium are trademarks of Intel Corporation in the United States, other countries, or both. UNIX is a registered trademark of The Open Group in the United States and other countries. SET and the SET Logo are trademarks owned by SET Secure Electronic Transaction LLC. Other company, product or service names may be trademarks or service marks of others. Information is provided "AS IS" without warranty of any kind. All customer examples described are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual environmental costs and performance characteristics may vary by customer. Information concerning non-IBM products was obtained from a supplier of these products, published announcement material, or other publicly available sources and does not constitute an endorsement of such products by IBM. Sources for non-IBM list prices and performance numbers are taken from publicly available information, including vendor announcements and vendor worldwide homepages. IBM has not tested these products and cannot confirm the accuracy of performance, capability, or any other claims related to non-IBM products. Questions on the capability of non-IBM products should be addressed to the supplier of those products. All statements regarding IBM future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only. Contact your local IBM office or IBM authorized reseller for the full text of the specific Statement of Direction. Some information addresses anticipated future capabilities. Such information is not intended as a definitive statement of a commitment to specific levels of performance, function or delivery schedules with respect to any future products. Such commitments are only made in IBM product announcements. The information is presented here to communicate IBM's current investment and development activities as a good faith effort to help with our customers' future planning. Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance that any user will experience will vary depending upon considerations such as the amount of multiprogramming in the user's job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve throughput or performance improvements equivalent to the ratios stated here. Photographs shown are of engineering prototypes. Changes may be incorporated in production models.