I've seen many figures banded around for the default I/O sizes for Advfs -
8k always seems to appear. In an attempt to get some kind of definitive
answer, I've decided to post this.
After looking at the chvol advfs command, it suggests that my domains have
default read/write sizes of 128 blocks (64k) - not 8k. So which is right?
Also with the consolidation mode set to on, does this mean that advfs will
queue I/Os up to the read/write size before issuing them? Therefore the I/O
size will always be constant?
For LSM, the kernel parameter volinfo.max_io is 128 (blocks I presume?).So
again LSM is using an I/O size of 64k by default.
All these parameters are tunable - within reason - so I can tweak them as
required, but I am right about the basics here?
Thanks,
Dave Campbell
(dave.campbell_at_vf.vodafone.co.uk)
Received on Tue May 25 1999 - 09:29:25 NZST