[1327] in SIPB-AFS-requests
Re: server migration
daemon@ATHENA.MIT.EDU (Marc Horowitz)
Sun Apr 17 03:08:46 1994
To: marthag@MIT.EDU
Cc: mhpower@MIT.EDU, ghudson@MIT.EDU, sorokin@MIT.EDU, sipb-afsreq@MIT.EDU,
charon-maintainers@MIT.EDU, rtfm-maintainers@MIT.EDU,
webmaster@MIT.EDU, usenet@MIT.EDU
In-Reply-To: Your message of Sun, 17 Apr 94 02:02:18 -0400.
<9404170602.AA10568@w20-575-50.MIT.EDU>
Date: Sun, 17 Apr 94 03:08:01 EDT
From: Marc Horowitz <marc@MIT.EDU>
>> Well, judging from the performance of noc.near.net (which, despite my
>> previous statement to the contrary, is an SS10), I'd say they aren't
>> twice as good as a Maxine. And an SS20 is not a large improvement over
>> an '10, it mainly only adds some more caching.
Well judging from the performance of marys.gza.com, a 128mb dual-proc
dual-SCSI bus SS10 can run circles around a maxine. Clearly, we need
more tests.
I'd have to think about Matt's proposal to have just one AFS server.
It's an interesting approach, but having two servers means a non-total
outage when if one server has to go down (expectedly or not). Perhaps
a compromise is to have two afs servers, but have the second only
serve "critical" volumes, like project.tcsh, and serve other things
(like discuss, maybe). This has security considerations which need to
be considered.
Marc
