Optimal use of many resource

Question

Hi Folks
My System:
OS: linux OEL 5.2 x86_64
Memory: 112GB
CPU: 4*6Core = 24Cores
Filesystem: ASM 11.1.0.7
RDBMS: Oracle 11.1.0.7 Archivelogmode
We have recently migrated an OLTP Database (~1.3TB) to this server above. It is a Single Instance DB and it is also the only one DB on this server no other apps. I have several question on how to set the parameters. This is the current configuration
memory_max_target                    big integer 95G
memory_target                        big integer 95G
sga_max_size                         big integer 65G
sga_target                           big integer 65G
pga_aggregate_target                 big integer 30GI set the sga_target to 65G because I cannot set the db_cache_size to a higher value than ~28G. I also set the log_buffer to 64MB but then I found the following Note on metalink:
The Log_buffer Default Size Cannot Be Reduced In 10g R2 ID: 351857.1
I set it to 64MB because the migration. I created a new empty 11g Instance and import with impdp over network the whole DB, with parallel index creation of huge tables.
Hence my Redo Buffer is ~ 212MB, a bit huge?
Oracle sets the memory setting to the following values:
Shared Pool      12800
Buffer Cache      48128
Large Pool      1024
Java Pool      3072
Other      1536
SQL> show sga
Total System Global Area 6.9482E+10 bytes
Fixed Size                  2161480 bytes
Variable Size            1.8790E+10 bytes
Database Buffers         5.0466E+10 bytes
Redo Buffers              223121408 bytesI have 6 Redolog groups each 2 members of 1GB a member. And there are 8 DBWR and 4 ARCH processes.
How would you use so much memory optimal? Is in your opinion the number of dbwr oversized? Have you other suggestion on how to use this resource optimal?
I think the "bottleneck" for this system will be the filesystem?
btw: it is not yet in production.
Thanks a lot
Regards Oviwan

Answer 1

I got it now. The target for the backupset is an nfs export (netapp filer) and I used native nfs (knfs) which eats up all resources. Now I switched to direct nfs, this is as fast as knfs and it don't need as much resources as knfs. The time for the backups are also equal with dnfs or knfs :)
but one thing is a bit crazy. on the filer I monitored the I/O: with knfs ~5000I/O and with dnfs ~500I/O --> 10times fewer I/O but same duration... nothing else was running on this filer during the tests...
mount options: rw,bg,hard,rsize=32768,wsize=32768,nfsvers=3,nointr,timeo=600,tcp,noac