From: Boris Z. (boris@conley.com)
Date: 02/24/99-08:42:19 AM Z
Message-ID: <01BE5FD9.F8AB7240.boris@conley.com> From: "Boris Z." <boris@conley.com> Subject: RE: Caching and callbacks Date: Wed, 24 Feb 1999 09:42:19 -0500 On Tuesday, February 23, 1999 11:15 PM, Mike Eisler [SMTP:mre@Eng.Sun.COM] wrote: > > I think the issue of callbacks may be a red herring that's diverting > > our attention from a more important problem: > ... On Wednesday, February 24, 1999 1:05 AM, Brent [SMTP:brent@Eng.Sun.COM] wrote: > Mike Eisler wrote: > > > Actually, I don't think the cache timeout GETATTRs are too much of > a problem. Currently they occur only if the attrs haven't been > refreshed in the last few seconds. So you'll see a GETATTR every > 3 - 30 sec. The problem with the file open GETATTR is that you > might see dozens per second. > Theoretically, NFS-3 client driver should send a GETATTR for every cached read or write request. This not only increases the network congestion, but also drastically reduces the local performance. Practically, clients 'cheat' to deal with this. I strongly believe that we have to include lease + call back oplock mechanism into V4. We all want to sleep well, right? :) Boris
This archive was generated by hypermail 2.1.2 : 03/04/05-01:46:43 AM Z CST