|
|
|
|
|
|
|
|
|
|
|
|
| home | help | back | first | fref | pref | prev | next | nref | lref | last | post |
Date: Mon, 18 Nov 91 15:14:23 -0500 From: "Jonathan I. Kamens" <jik@pit-manager.mit.edu> To: henry@ads.com (Henry Mensch) Cc: gmc@helix.nih.gov, info-afs@transarc.com In-Reply-To: [448] From: henry@ads.com (Henry Mensch) Date: Mon, 18 Nov 91 09:57:47 -0800 Reply-To: henry@ads.com is there any reason to *not* have a server also be a client? (doesn't this sound too much like the hair club for men: "i'm not only a server ... i'm a client!") We have noticed severe performance degradation when writing to a volume that is stored on a machine from that same machine. I don't know whether this is just us, or a general problem that is known to Transarc. It would seem to me that if the Kernel has to context switch over and over between the AFS client processes and the AFS server processes, it's going to get a bit bogged down.... Jonathan Kamens jik@MIT.Edu MIT Information Systems/Athena
|
|
|
|
|
|
|
|
|
|
|
|
| home | help | back | first | fref | pref | prev | next | nref | lref | last | post |