On Mon, 11 Oct 2004 15:34:36 -0700, Johnny Walton <[EMAIL PROTECTED]> wrote:

> I can't seem to remember what all these entries mean.

http://lxr.linux.no/source/Documentation/sysctl/vm.txt

The 7th and 8th parameter seem to define the percentage of dirty that
will kick in bdflush synchronously. The default for these is 60% and
20% so you want those to go down.

We must not forget Marcy's warning about the 7700E. You do want to
measure the I/O response of your DASD subsystem. The pure device
response time may not be too helpful if you don't know how big the I/O
is. Fortunately with ESCON this should be fairly easy: the connect
time lets you compute the number of blocks per I/O.
>From another issue I did learn that average I/O response has its
limitations (e.g. when 1% of the I/O's takes 100 times as long - does
not show in the average but is bad).

>  Thanks for sticking with me on this.
My pleasure. It's much better to work on someone else's problem than
to work on your own problems... ;-)

> Could you email me a signed version;)
Promised.. once we have sorted out this problem of yours.

--
Rob van der Heij                  rvdheij @ gmail.com

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to