What is the measure for tablespace IO writes efficiency?
We've started transferring in the last days our storage from DMX3 to Vmax.
I want to compare performance of specific tablespace before and after the transferring.
In AWR ,in the section of Tablespace IO stats I checked the measure "Av Rd(ms)" regarding reads.
Is there a measure for writes?
How can I know that I've got an improvements regarding writes?
Provided timed_statistics is set to TRUE v$filestat shows the information you are looking for.
Having said that, please take into consideration that Write I/O performance is a tricky matter in the sense that:
1- Writes to datafiles are done by DBWR in the background so as long as DBWR can keep up with the load you really don't care about I/O performance;; DBWR will get it done and user community will never notice performance fluctuations.
2- Writes to logs and temporary segments on the other hand are done in the foreground therefore poor performance in such segments will immediately have a negative impact on overall performance.
Bottom line, don't waste your time analyzing Write I/O on standard "data and indexes" data files; focus your attention on Logs or Temporary segment data files.
Last but not least, when looking at Write I/O I would suggest to check what's the performance your I/O subsystem is giving you - talk to your storage guy and ask her/him to report 'service levels" to you. Also ask for the technical specs of your particular I/O subsystem so to compare what the vendor sold against actual service levels.
Disclaimer: Advice is provided to the best of my knowledge but no implicit or explicit warranties are provided. Since the advisor explicitly encourages testing any and all suggestions on a test non-production environment advisor should not held liable or responsible for any actions taken based on the given advice.