I recently investigated an IO performance “spike” on a large 220.127.116.11 transactional system and I thought I would cover some interesting issues found. I am going to take the approach of detailing the observations made from our production and test systems and avoid attempting to cover how other versions of Oracle behave. The investigation also uncovers a confusing database statistic which we are currently discussing with Oracle Development so they can decide if this is an Oracle coding bug or a documentation issue.
The initial IO issue
We run a simple home grown database monitor which watches database wait events and sends an email alert if it detects either a single session waiting on a non-idle wait for a long time or the total number of database sessions concurrently waiting goes above a defined threshold. The monitor can give a number of false alerts but can also draw our attention to some more interesting events.
One day the monitor alerted me to two separate databases waiting on “log file sync” waits at exactly the same few seconds…
View original post 1,536 more words