Date: Sat, 13 Mar 1999 17:22:06 -0600 (CST) From: James Wyatt <jwyatt@RWSystems.net> To: freebsd-isp@FreeBSD.ORG Cc: Mark Mayo <mark@vmunix.com>, Gary Palmer <gpalmer@FreeBSD.ORG> Subject: Re: Mail server setup Message-ID: <Pine.BSF.4.05.9903131657220.29916-100000@kasie.rwsystems.net> In-Reply-To: <89008.921212284@gjp.erols.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 11 Mar 1999, Gary Palmer wrote:
> Anyone who buys storage direct from Sun needs their head looked at. The A3500
> is a rebadged unit from Symbios (now LSI Logic), which we get for a lot less
> than Sun charge for it. The A5000 is JBOD on fiberchannel. No write cache.
> Interesting unit. Sucks for real applications. Again, LSI have a real
> fiberchannel based RAID system, and are trying their hardest to make it
> interoperate with other equipment enough (e.g. FC switches) that it may
> actually work, unlike Suns stuff.
I've had great success with IBMs SSA setup, we currently have 16 4.5G
drives in one drawer and 32 9GB in another two running HA/CMP for a
250,000 message/day EDI system. I was *amazed* though, when I saw the
costs for our next drawer: $8500 for the drawer and $3500/drive for the 16
18GB drives! (I intend to try the RAID controller this time because direct
mirroring is so costly.) The faceplate for the panel is $170! The new DLT
library was equally shocking.
Our customer wants to pull-up *any* EDI message over 7 years of traffic
with numerous keys within the data (Car ID, VIN, Bill No, etc...).
Fortunately they have the pockets to pay for it when they demand 100%
functionality up 99%+ of the time.
These units *do* work very well, though, and I can stretch SSA cables
hundreds of feet when the server room is short on space. We use HA which
allows us to rapidly failover machines and that is harder to scale with
plain Diff-SCSI. One of our tricker systems management issues was telling
when the failover happens. We solved it for looking at the backup LAN
cards which go away when HA has tripped.
Is there anything like this for FreeBSD? I know there are SCSI arrays, but
does anything allow for failover? We do *NOT* use NFS for any part of our
system but user home directories. (We tried it on an older remote video
imaging system, but NFS mounts took way too long to force-out)
The thing I like about directing proxies is that they can be in parallel
for SPoF reduction. If you have fractional mail spools, only a part of the
user base is affected, but it will always be those folks that need it
instantly and don't read advisories. 8{( The simpler function the node has
(like only a-k usernames), the more quickly your outage will be over.
Like others have said: enough rambling, I gotta get back to work - Jy@
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.05.9903131657220.29916-100000>