On Thu, Dec 14, 2000 at 04:48:32PM -0800, Henry B. Hotz wrote:
> >
> >This doesn't explain why raw thoughput of TCP mounts is much higther than
> >UDP (A solaris client writes at 7MB/s to my NetBSD server with UDP mounts,
> >and 11MB/s with TCP).
> 
> TCP can play games with ack's to compensate for latency.  UDP can't 
> unless the NFS code duplicates that low-level optimization (which I 
> doubt).  I would expect large-block operations to be faster over TCP 
> if the blocks are larger than the fragment size for the connection in 
> use.

Yes, that should be it. The speed improvement was for large files read/write.
For directory operations (lots of create/delete, etc ...) TCP or UDP
didn't show significant differences.

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--