>>>
>>>Our central NAS system has performance issues with nfs on
>>>solaris.

That is confusing. This is a suse list. Is the NAS actually
just a Solaris server exporting filesystem? If so, that is
not OSCP, but for testing and backups, I'm sure that is
just fine.

When writing/reading over a gbit-link we see "only"
>>>35 mb/s within a single thread. When adding a second thread
>>>(or process) the throughput increases - up to 80 mb/s with 4
>>>threads (or processes).

You didn't say what the NAS device is. Getting 80MB/s
of writes to a typical filer is actually really really good.
Generally they bottlenck due to NVRAM cache checkpointing
before that (but that does depend on the size of write too).

>>>
>>>However when writing data to disk with lets say 4 dbw
>>>processes we can max out the 80 mb/s. But how does reading
>>>from disk relates to that ? Lets say we do a full table scan
>>>in a single session and therefore have to read the whole
>>>table - does oracle use more than one process (or thread) to
>>>do that (remember my problem: one session or thread can only
>>>read 35 mb/s at maximum speed) ?

Oracle will use more than one scanning process if PQO gets
fired off, otherwise no. There was no mention of
what filesystemio_options is set to either.



--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting