Subject: Re: interfaces, receive buffers, and NFS
To: John Hawkinson <jhawk@MIT.EDU>
From: Charles M. Hannum <email@example.com>
Date: 04/06/2001 09:40:34
On Tue, Apr 03, 2001 at 09:16:08AM -0400, John Hawkinson wrote:
> I don't think that the decision to raise the nfs read/write size
> from 8k to 32k should be governed exclusively by host-side
> buffer issues.
Nor is buffer size even a useful metric. There are many devices with
which we can receive at line rate even though they have small buffers.
In fact, we can do this with *most* Ethernet cards. It's only garbage
like the 3c509 that actually has problems.
> For NFS in environments where it crosses a router, or [help us!]
> a WAN, a 32k fragmented UDP packet can be far more severe than
> an 8k one, because it may be substantially more likely
> that the drop of a single fragment will occur, for two reasons:
In this case, you should be using NFS over TCP.