It’s rare to find professional level tools that are written by the people who use them. Most professional tools are written by teams of people who have little idea of how they are used and many users write tools that never reach a professional level. Thus it is a pleasure to see the tools put out by Method R with Cary Millsap and Jeff Holt. Jeff and Cary have been well known in the industry for their work and contributions to the world of Oracle performance tuning. With Method R tools they not only share their know how and tools they use to tune systems, but they have polished them into solid, clean , tight professional level tools that are dependable, high quality and written with integrity.
MR Trace ($49.95) – automatically creates, collects and manages trace files for queries run in Oracle’s SQL Developer using extended tracefile options targeted at collecting pertinent performance information for optimizing SQL statements.
mrls – trace file lister, examples, list trace files sorted by LIO or sorted by rows returned
mrcallrm – trace file cropper – retaining required info and filtering undesired
Trace file timings:
mrtim – timestamp converter
mrtimfix – your trace file tim value fixer, fix 11gR1 bug in trace files & change timing to UNIX epoch
mrnl – pre-pend lines in trace with line #, time, duration and other options as well
Is like the gold standard for the Method R tools . The profiler takes all the know how and analysis and wraps it up in a clean crisp html UI that makes for easy browsing , exploration and analysis
I look forward to blogging soon on some of the innovative and powerful features of Method R Profiler
A quick summary of things in the report
1. Task-Level Profiles 1.1 Profile by subroutine subroutines: events like: db file sequential read db file scattered read > spark line table , each line is a bucket between 1us - 1000 secs, orders of 10 call count total seconds 2.2 next subroutine 3. Contribution to Subroutine by Statement 3.1 Statement Contributions to <> statement text (with sql hash) seconds duration count mean min max drill down to stats 3.2 next subroutine 4. Statement Detail 4.1 Statement Report for <> Oracle hash value: Full statement text: drilldown Re-use: # distinct text Text size: x lines, y bytes # of distinct plans: X Optimizer goals: response time contribution: 4.1.1 Profile by Contribution to Parent Cursor 4.1.2 Profile by Subroutine waits, CPU, time by seconds, count, mean, min, max, skew sparkline 4.1.3 Profile by Database Call for FETCH beween calls EXEC PARSE cursor close show seconds CPU rwos processed bufer cache access total CR CU PIOs Library cache misses 4.1.4 Execution Plans and Placeholder Values Execution plan seconds for row ! seconds including descendents rows returned CR LIOs PIOs read PIOs writes 4.1.5 Statement Text 4.2 ... next statement ... 5. Configuration information