NFS read size nfs:nfs3_bsize
2007-12-24 21:45:00
I'm doing some quite large copies (about 300G) via nfs on a regular basis,
and I was trying to increase the rsize parameter past the solaris nfs3
default of 32K to improve performance. Unfortunately the 32K limit is set by
a couple of kernel parameters. Looking at the kernel tuning document, the
notes for nfs:nfs3_bsize say "Do not set this parameter too high because it
might cause the system to hang waiting for memory allocations to be
granted."
My question: Exactly what is defined as "too high"? 64K? 512K?
This is also very hard for me to test, as while I have some ultra-5's I can
play with, there is no way I get them under the same kind of general load as
the E450 and E420r which are the production system, which means I don't feel
that test results on the play machines are valid for the production servers.
Does anyone have any experience in this area?
Thanks,
Tristan.
Comments
Got something to say?
You must be logged in to post a comment.
