Confusion over AWR report interpretation
Hi,
While analyzing some reported performance issue, I came across the following situation:
One box (SunOS Sparc) holds two database server. The server has 64GB RAM and 16 Logical CPUs. Say the SIDs of the DB servers are A & B. Performance issue has been reported with DB A. Reports/procedures taking huge time to complete at DB A.
As part of analysis we have generated AWRs during the time of reported performance issue. Within the Top 5 wait events following wait events found most of the time:
1. log file sync (Avg Wait(ms) 80-120)
2. log buffer space (Avg Wait(ms) 600-800)
3. log file parallel write (Avg Wait(ms) 400-500)
These are the events and values for database A. Apart from that the Av Rd (ms) part of IO Stats some time reflecting values more than 100ms for some datafiles.
As we are in communication with client, it has been incidentally observed that the other database (DB B) has same events with similar values. IO Stats of both the databases are also same at certain intervals.
But it is to be noted that the users of DB B are not experiencing any kind of performance issues. Data volume of DB B is more than double than that of DB A.
I am not able to understand that with similar kind of wait events and IO Stats, how could the user experience be different for both the databases.
Database Version: 10.2.0.5 64 bit
Thanks in advance for your valuable time.
Regards,
Joy