[4899] in linux-scsi channel archive
Re: Quick overview of Disk Sharing (Was: Re: dual-attached disk)
daemon@ATHENA.MIT.EDU (Mr M S Aitchison)
Thu Oct 15 09:18:21 1998
Date: Thu, 15 Oct 1998 12:00:48 +1300
From: physmsa@cantua.canterbury.ac.nz (Mr M S Aitchison)
To: linux-scsi@vger.rutgers.edu
"Failover" type of sharing has been around for many years - e.g. the
Data General Nova/Eclipse line around the late 1970's with their IPB
and watchdog timer. And modern systems (including DG again, but also
many disk array subsystems such as from Infortrend) can have two
computers connected to disks, one ready to take over in all respects
(control the disk, assume the IP number) if the other fails.
I think the hardware that makes it easy is a RAID controller with
enough channels (or whatever the jargon) to connect to two computers on
different SCSI cables.
I also recall hearing of disk subsystems that not only can connect to
multiple computers but to the ethernet directly - but I only have fuzzy
ideas of what they claimed could be done simultaneously.
However there a couple of totally different approachs, such as used by
Sun with their cached filesystem and automounts. One is to let a
server use some of its own local disk to cache a network server - it
goes to its copy often enough for reads to improve efficiency, but
ultimately it has to keep checking on the original server so if that
goes down your second server will stop sooner or later (sooner for
writes, of course). A second idea is to have two servers with two
disks, and clients automount whatever responds first. Again this is
pretty much only good for increasing reliability of read-only
partitions (the two servers have to synchronise their copies somehow).
And it suffers in the Sun implementation from having to stick with one
server for a relatively long time - if that server dies during that
time you are very stuck.
I think Linux could take both ideas, plus the idea of RAID or logging
changes/journalling, and implement something more reliable but still
aimed at systems doing many more reads than writes. But to do justice
to this it probably needs a lot more than just a beef-up of an
automounter but something like IBM?/Transarc Distributed File System
(I've had a quick look at:
www.transarc.com/dfs/public/www/htdocs/.hosts/external/Product/EFS/DFS/dfs_broc.htm
and see it mentions "UNIX" but I don't know if there is a Linux
version).
---------------------------------------------------------------------------
Mark Aitchison, \_ Phone: +64 3 364-2947 home 337-1225
Dept of Physics & Astronomy, <| Fax: +64 3 364-2469 or 364-2999
University of Canterbury, /) Web: www.phys.canterbury.ac.nz/~physmsa
Christchurch, NEW ZEALAND. (/' E-mail: M.Aitchison@phys.canterbury.ac.nz
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
