NO FILE SYSTEM CACHING - Automatic?

[Patrick Finnegan]

The blurb at this link seems to suggest that concurrent i/o can be
enabled on AIX at table space level by setting "no file system
caching" on the table space and nothing else. Can somebody confirm
that this is true and that Concurrent i/o does not have to be enabled
as a file system mount option.

We are talking DB2 8.2.8.

http://publib.boulder.ibm.com/infoce...c/t0023622.htm
Thanks.

[Patrick Finnegan]

On Oct 21, 7:46*pm, Patrick Finnegan <[e-mail]>
wrote:

The blurb at this link seems to suggest that concurrent i/o can be
enabled on AIX at table space level by setting "no file system
caching" on the table space and nothing else. *Can somebody confirm
that this is true and that Concurrent i/o does not have to be enabled
as a file system mount option.

We are talking DB2 8.2.8.

http://publib.boulder.ibm.com/infoce...ic/com.ibm.db2....

Thanks.

And what about cio for transaction logs? Has anyone had good
experience with that?

http://publib.boulder.ibm.com/infoce...c/t0025268.htm

[Patrick Finnegan]

On Oct 21, 6:46*pm, Patrick Finnegan <[e-mail]>
wrote:

The blurb at this link seems to suggest that concurrent i/o can be
enabled on AIX at table space level by setting "no file system
caching" on the table space and nothing else. *Can somebody confirm
that this is true and that Concurrent i/o does not have to be enabled
as a file system mount option.

We are talking DB2 8.2.8.

http://publib.boulder.ibm.com/infoce...ic/com.ibm.db2....

Thanks.

Regarding the above PMR, you are absolutely correct, CIO doesn't have
to be enabled at the filesystem level (mount -o cio). When we set a
'no file system caching' policy we make a specific OS call which
utilises CIO irrespective of the mount.

In fact we prefer you do it this way because by setting CIO as a mount
option all files are CIO enabled, which may not be benefitial to some
OS files.