ActiveBase Tuning Robot automatically collects problematic SQL queries from AWR, analyzes them using hints to find alternative execution plans, benchmarks the alternatives, and outputs tuning rules to improve performance without changing application code. It identifies top SQL, tunes them in parallel, and tests improvements in pre-production. Configuration involves settings for SQL classification, parallel tuning, AWR data collection, and benchmarking options. The Tuning Robot provides continuous, hands-free application optimization and performance improvements.
Rising Above_ Dubai Floods and the Fortitude of Dubai International Airport.pdf
Tuning Robot Quick Tour
1. ActiveBase Ltd. All Rights reserved ActiveBase Tuning Robot TM Quick Tour Learn how ActiveBase Tuning Robot TM expands ActiveBase SQL Expert TM functionality with an automatic AWR TOP-SQL collector and SQL benchmark scheduling capabilities.
2. Introduction to ActiveBase Tuning Robot TM > ActiveBase Tuning Robot TM software delivers a continuous optimization of your Oracle applications, saving time and expert resources. > Installed on a server, it automatically retrieves heavy ‘Select’ SQL requests from AWR (collector module), analyzes them using various Oracle ‘Hints’ (guaranteeing result set) and benchmarks them -> highlighting the best alternative. > Rule.xml file is automatically created for import into ActiveBase Performance TM , applying the improvements without touching application source-code or databases. It enables to verify improvements in pre-production and/or in production (when code fixing is not feasible) - resulting in x10-100 response time improvements. ActiveBase Ltd. All Rights reserved
3.
4.
5. How the Tuning Robot was configured: SQL classification and Parallel executions In an application tuning assignment, three parallel Tuning Robot batches where executed: Batch 1: Long running SQL requests with average elapse time > 10 sec. Batch 2: Medium running SQL requests with average elapse time between 1 – 10 seconds using a high parallelism degree to gain quick optimizations, where alternatives were compared based on elapse of 5 serial executions Batch 3: Short running SQL requests with average < 1 sec., compared based on elapse of 100 serial executions ActiveBase Ltd. All Rights reserved
6. Tuning Robot configuration Tuning Robot requires configuring two files: DB.Properties - defining analysis and benchmark options AWR.Properties - setup collector for collecting AWR statistics ActiveBase Ltd. All Rights reserved
7. DB.Properties parameter settings > maxThreads=Number of parallel statement optimizations (e.g., maxThreads=2 – 2 threads are tuning two statements in parallel). > maxRunningTime=Total tuning process elapse time. > analyzeLevel=Defines the number of hint combination investigated on the SQL statement. > maxAlternatives=Total amount of alternatives with unique execution plans analyzed > autoCancelPercent= automatically cancelling alternatives with execution time > X% from the best so far. > benchmarkOptions.executionsNumber = Execute each alternative x times for accurate execution statistics > sessionParameters = define ‘Alter session’ session parameters ActiveBase Ltd. All Rights reserved
8. AWR.Properties parameter settings > jdbc.url= AWR statistics can be retrieved from production while tuned in pre-production. > time.start and time.end =define relevant time slice in the AWR > elapse.min and elapse.max = AWR statements running over x second and under y seconds > elapse.top= AWR top z statements > test.xml= name of the XML file containing the rules to be imported > sessionParameters = define ‘Alter session’ session parameters ActiveBase Ltd. All Rights reserved