Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: MASSIVE degradation of insert using sequence ID's via JDBC on 9i?
Dear all;
Thanks for all your help and pointers.
I have still not resolved why I'm cripping
9i, but I have learnt about trace events
and analysing trace files.
I have now run my test (as listed elsewhere)
against both 9i and 10g, with
10046 tracing at level 12.
As well as tkprof, I used the
AppsDBA resource profiler
http://home.comcast.net/~arivenes/scripts/resource.zip
to analyse the results.
The biggest (and it's BIG) difference is clearly shown in the summary:
10 looks "ok"
INTERVAL RATES AND WAITS (service time + wait time)
Total Event % Event Total --------------- Cumulative ---------------- Event Name Time (sec) Time Events Avg Time(sec) Min Time(sec) Max Time(sec)
-------------------------------- ----------------- ------- ---------- ------------- ------------- -------------
unaccounted-for 5.871474 33.73 SQL*Net message from client 5.448164 31.30 20,007 0.000272 0.000164 0.197724 CPU Service 4.706311 27.04 25,974 0.000181 0.000000 0.097985 log file sync 1.231318 7.07 5,097 0.000242 0.000004 0.056876 SQL*Net message to client 0.088047 0.51 20,007 0.000004 0.000002 0..000131 enq: RO - fast object reuse 0.050858 0.29 4 0.012715 0.000007 0.048422 log file switch completion 0.009098 0.05 1 0.009098 0.009098 0.009098 rdbms ipc reply 0.001343 0.01 4 0.000336 0.000126 0.000847
-------------------------------- ----------------- ------- ---------- ------------- ------------- -------------
Total 17.406613 100.00 71,094
While 9 appears "unwell"
INTERVAL RATES AND WAITS (service time + wait time)
Total Event % Event Total --------------- Cumulative ---------------- Event Name Time (sec) Time Events Avg Time(sec) Min Time(sec) Max Time(sec)
-------------------------------- ----------------- ------- ---------- ------------- ------------- -------------
log file sync 572.589406 97.21 5,058 0.113205 0.000043 1.000852 unaccounted-for 6.100452 1.04 SQL*Net message from client 5.280490 0.90 20,007 0.000264 0.000146 0.133374 CPU Service 4.710000 0.80 26,420 0.000178 0.000000 0.100000 db file sequential read 0.292716 0.05 66 0.004435 0.000025 0.016963 SQL*Net message to client 0.032749 0.01 20,007 0.000002 0.000000 0.000722 rdbms ipc reply 0.012621 0.00 8 0.001578 0.000810 0.003936
-------------------------------- ----------------- ------- ---------- ------------- ------------- -------------
Total 589.018434 100.00 71,566
the log file sync event count is very similar, but where 10g is doing each flush in 0.000242, 9i takes a rather leisurely 0.113205.
Googling shows that "log file sync" simply mean the LGWR system saving redo information.
I'm off to learn more, but would like to say "thank you" to all the good folks here who helped me get this far.
BugBear (still happy to receive advice) Received on Wed Jun 29 2005 - 08:55:40 CDT