On Fri, 2 Apr 1999, Brian Pawlowski wrote:
I am in the process of writing this all up.
Setting options nfs.udp.xfersize 32768 helped a lot, now were seeing ~25 MB/sec transfers to and from the filer. Why isn't this the default?
In some sick twisted way I remembered a quote I once read in an OS design book. I cite:
A write system call may be broken down into several RPC writes, because each NFS write or read can contain up to 8K of data and UDP packets are limited to 1500 bytes.
The author was obviously assuming Ethernet medium when he stated that UDP packets can be at most 1500 bytes, but, more importantly, the 8K limit probably came from some earlier NFS or RPC standard. I wouldn't be surprised if someone at NAC used that as a safe default.
Tom