DB2 SQL Tuning Tips for z/OS Developers

Tony Andrews has more than 23 years’ experience in the development of IBM DB2 relational database applications. Most of this time, he has provided development and consulting services to Fortune 500 companies and government agencies. Tony has written literally thousands of queries and programs during his development years, and he has also served as a DB2 database analyst. For the past 10 years, Tony has been splitting his time between consulting engagements and training. His main focus is to teach today’s developers the ways of RDMS application design, development, and SQL programming— always with a special emphasis on improving performance. Tony’s training, consulting, and speaking engagements are through his employer, Themis, Inc., an onsite and virtual instructor-led, hands-on IT training company recognized internationally. It offers more than 400 IT courses and helps to support International DB2 Users Group North America (IDUG NA) and Europe, Middle East, and Africa (IDUG EMEA), along with many DB2 user groups. Tony is a current IBM champion and regular lecturer at industry conferences and local user groups. You may have seen him present at such events as IDUG NA and EMEA. He is well known for his “Top 25+ Tuning Tips for Developers” presentation. Tony graduated from Ohio State University with a major in business and a minor in mathematical statistics. He currently resides in Dublin, Ohio.

• CHAPTER 1  SQL Optimization Top 100+     1 1. Take Out Any/All Scalar Functions Coded on Columns in Predicates     2 2. Take Out Any/All Mathematics Coded on Columns in Predicates     3 3. Code Only the Columns Needed in the Select Portion of the SQL Statement     4 4. Stay Away from Distinct if Possible     4 5. Try Rewriting an In Subquery as an Exists Subquery     5 6. Always Make Sure Host Variables Are Defined to Match the Columns Datatype     6 7. Because Or Logic Can Be Problematic to the Optimizer, Try a Different Rewrite     6 8. Make Sure the Data Distribution and Other Statistics Are Good and Current in the Tables Being Processed     8 9. Use UNION ALL in Place of UNION Where Possible     9 10. Consider Hardcoding Versus Using a Host Variable     9 11. Minimize DB2’s SQL Requests     11 12. Try Rewriting Range Predicates as Between Predicates     12 13. Consider Using Global Temporary Tables     13 14. Give Prominence to Stage 1 over Stage 2 Predicates     14 15. Remember That the Order of (Some) Predicates Does Matter     15 16. Streamline Multiple Subqueries     16 17. Index Correlated Subqueries     17 18. Get to Know the DB2 Explain Tool     17 19. Use Tools for Monitoring     18 20. Employ Commit and Restart Strategies     19 21. Implement Good Index Design     19 22. Avoid Discrepancies with Non-Column Expressions     20 23. Begin with All Filtering Logic Outside Application Code     21 24. Ensure That Subquery Predicates Involving Min and Max Have the Possibility of Nulls Being Returned Handled     21 25. Always Code For Fetch Only or For Read Only with Cursor Processing When a Query Is Only Selecting Data     22 26. Avoid Selecting a Row from a Table to Help Decide Whether the Logic in the Code Should Execute an Update or an Insert     23 27. Avoid Selecting a Row from a Table in Order to Get Values for an Update     23 28. Make Use of Dynamic SQL Statement Caching     23 29. Avoid Using Select *     24 30. Watch Out for Nullable Columns or Times When SQL Statements Could Have Nulls Returned from the Database Manager     25 31. Minimize the Number of Times Open and Close Cursors Are Executed     25 32. Avoid Not Logic in SQL     26 33. Use Correlation IDs for Better Readability     26 34. Keep Table and Index Files Healthy and Organized     27 35. Take Advantage of Update Where Current of Cursor and Delete Where Current of Cursor     27 36. When Using Cursors, Use ROWSET Positioning and Fetching Using Multiple-Row Fetch, Multiple-Row Update, and Multiple-Row Insert     28 37. Know the Locking Isolation Levels     28 38. Know Null Processing     30 39. Always Program with Performance in Mind     31 40. Let SQL Do the Work     32 41. Code with Lock Table     32 42. Consider OLTP Front-End Processing     33 43. Consider Using Dynamic Scrollable Cursors     34 44. Take Advantage of Materialized Query Tables to Improve Response Time (Dynamic SQL Only)     35 45. Insert with Select     37 46. Take Advantage of Multiple-Row Fetch     38 47. Take Advantage of Multiple-Row Insert     39 48. Take Advantage of Multiple-Row Update     40 49. Take Advantage of Multiple-Row Delete     42 50. Try Scalar Fullselects Within the Select Clause     42 51. Take Advantage of REOPT ONCE and REOPT AUTO in Dynamic SQL and REOPT VARS and REOPT ALWAYS in Static SQL     43 52. Identify Times for Volatile Tables     44 53. Use the ON COMMIT DROP Enhancement     45 54. Use Multiple Distincts     45 55. Take Advantage of Backward Index Scanning     46 56. Watch Out for the Like Statement     46 57. Set Your Clustering Index Correctly     47 58. Use Group By Expressions if Needed     48 59. Watch Out for Tablespace Scans     48 60. Do Not Ask for What You Already Know     49 61. Watch the Order of Tables in a Query     49 62. Use Left Outer Joins Over Right Outer Joins     51 63. Check for Non-Existence     51 64. Use Stored Procedures     52 65. Do Not Select a Column in Order to Sort on It     53 66. Always Limit the Result Set if Possible     53 67. Take Advantage of DB2 V8 Enhanced DISCARD Capabilities When It Comes to Mass Deletes     54 68. Take Advantage of the DB2 LOAD Utility for Mass Inserts     54 69. Watch Out for Materialization of Views, Nested Table Expressions, and Common Table Expressions     55 70. Consider Compressing Data     56 71. Consider Parallelism     57 72. Keep the STDDEV, STDDEV_SAMP, VAR, and VAR_SAMP Functions Apart from Other Functions     58 73. Consider Direct Row Access Using ROWID Datatype (V8) or RID Function (V9)     58 74. Test Your Queries with Realistic Statistics and a Level of Data to Reflect Performance Issues     60 75. Specify the Leading Index Columns in WHERE Clauses     61 76. Use WHERE Instead of HAVING for Filtering Whenever Possible     62 77. Keep in Mind Index Only Processing Whenever Possible     62 78. Index on Expression in DB2 V9     63 79. Consider the DB2 V9 Truncate Statement     64 80. Use DB2 V9 Fetch First and Order by Within Subqueries     65 81. Take Advantage of DB2 V9 Optimistic Locking     65 82. Use the DB2 V9 MERGE Statement     66 83. Understand the DB2 NOFOR Precompile Option     68 84. Consider Select Into Using Order By     69 85. Code Boolean Term Predicates Whenever Possible     69 86. Try Transitive Closure Coding    70 87. Avoid Sorts with Order By    71 88. Use Joins Instead of Subqueries Whenever Possible    71 89. Watch Out for Case Logic    71 90. Take Advantage of Functions in the Order By Clause    72 91. Know Your Version of DB2     72 92. Understand Date Arithmetic     73 93. Know Your High-Volume Insert Choices     73 94. Know About Skip Locked Data (V9) for Lock Avoidance. . . . . .75 95. Sort Your Input Streams     75 96. If You Need True Uniqueness, Try the V8 Generate_Unique Function     76 97. Know the New Options for Declared Temporary Tables     76 98. Watch Out When Executing Get Diagnostics     77 99. Order Your In List Appropriately     77 100. Update and Delete with Select (V9)     77 101. Execute SQL Statements Only if Necessary     78 102. Take Advantage of In-Memory Tables     78 103. Stay Away from Catchall SQL Statements     79 104. Avoid Unnecessary Sorting     79 105. Understand Expressions and Column Functions     79 106. Watch Out When Combining Predicates     80 107. Add Redundant Predicates to Search Queries     80 108. Take Advantage of Improved Dynamic Caching (V10)     81 109. Try Currently Committed for Lock Avoidance (V10)     82 110. Try System Temporal Tables for Historical Data (V10)     83 111. Try Business Temporal Tables for Historical Data (V10)     85 112. Know Your Ranking Functions (V10)     86 113. Take Advantage of Extended Indicators (V10)     87 114. Get Greater Timestamp Precision (V10)     88 115. Try Index Includes (V10)     89 116. Use With Return to Client (V10)     89 CHAPTER 2  DB2 SQL Hints     91 1. Try the Optimize for 1 Row Statement at the End of the SQL Statement     91 2. Add the A.PKEY = A.PKEY Predicate to the SQL Query, Where PKEY Equals the Primary Key Column of the Table     92 3. Disqualify an Index Choice     93 4. Change the Order of Table Processing     95 5. Use Distributed Dynamic SQL     96 CHAPTER 3  SQL Standards and Guidelines     99 For COBOL Developers     99 For All SQL Developers     102 CHAPTER 4  SQL Program Walkthroughs     107 CHAPTER 5  Existence Checking     111 Example 1     111 Example 2     113 CHAPTER 6  Runstats     115 CHAPTER 7  Initial Steps in Tuning a Query     117 APPENDIX A  Predicate Rewrite Examples     121 Predicate Rewrites: Transitive Closure     122 APPENDIX B  DB2 SQL Terminology     125 Index     131