Ankit Ashok Aggarwal wrote:check OS log file
Question is - which method you are using to check datafile I/O errors ?
AND how to resolve high physical reads and writes??at what value does reads go from being OK to being "high"?
Ankit Ashok Aggarwal wrote:I dont think that there is any possibility to mention that how much of an IO is going to be good or bad for a db in general. The reason is this is not quantifiable . For example, in a store, may be on one day no one enters and on another day, there is no place to stand(because there was a sale announced by the store) . So now, how would you say which one was not needed? Rather than looking for some best practice values, it's better that you keep a track of the normal working of your db and when you see an exceptional high in the values of the user IO events like DB File Scattered Read , DB File Parallel Read etc and when that happens, you should check back the work that you are doing and see if that can be tuned by some way.
AWR do project i/o stat in terms of segements,query and etc.
here my concern is what should be ideal threashold value for a DBA to act on high physical reads/writes?
and how he can address this problem if it is affecting database performance??