Subject: Re: What kind of storage tray is a sun 540-3234 or 540-3005
To: Johan A. van Zanten <johan@brahma.giantfoo.org>
From: Matthew Jacob <mjacob@feral.com>
List: port-sparc
Date: 08/12/2003 09:13:58
SES should also work with them.
On Tue, 5 Aug 2003, Johan A. van Zanten wrote:
>
> ---In message <26.3d1f22cb.2c6141c0@aol.com>
> >I picked up a nice rack mass storage array with a bunch of 9g
> >drives in it. The rack is powered by a differential controller that
> >runs to three boxes of drives that are 68 pin scsi (non-differential).
> >I would like very much to run one of the drive trays externally on
> >my Ultra 1. I can't seem to find any model ID info on the mass
> >storage cabinet or the individual drive trays, so I can dig up some
> >docs on the critters. The only numbers I can find are:
> >
> > 540-3005
> >
> >and
> >
> > 540-3234
> >
> >each holding 7 drive boxes, 540-2942. It appears they are wired
> >for 240v, and I need to find out if they can be switched to 110v.
> >The label on the back has a possible model number OSM-3000-02
> >with a part number 501-8745.
> >
> >None of these numbers seems to key out on a search.
> >
> >Any insight as to what they are, and where some docs might be,
> >is appreciated.
>
>
> I think that might be an RSM Array 2000, which was later renamed to
> "A3000" and "A3500".
>
> http://docs.sun.com/db/doc/805-2782-10
> http://sunsolve.sun.com/handbook_pub/Systems/RSM2000/RSM2000_Rack.html
> http://sunsolve.sun.com/handbook_pub/Devices/SCSI/SCSI_RSM_2000_Ctlr.html
>
>
> It's a (hardware) RAID array. Intel processors on the controllers,
> actually. Very resilient; nearly everything is hot-swap and redundant.
> Can have two redundant, load-balancing, controllers.
>
> The disks are pretty straight forward SCSI. Those trays were used in a
> number of different arrays.
>
> The controllers came with write cache that was battery-backed, and if
> memory serves, the controllers will log errors and disable the cache if
> the battery is dead, or past a certain age.
>
> There's a software package for managing the controllers over the SCSI
> connection. The executable was called "rm6".
>
> It was a pretty complex driver; to take advantage of dual-SCSI pathing to
> the controllers, one talked to the devices via a tree of links in /dev/osa
> (i think.. it's been awhile) that enable the driver to automatically
> fail-over to the second SCSI channel to the controllers if there was a
> problem with the primary channel.
>
>
> I liked those arrays; they were good.
>
> I would think you would be able to connect a drive tray to an ultra
> no-problem, and treat them as the JBOD they are.
>
> The trays are often referred to as "RSM trays". They were also sold with
> the SPARCStorage Array 2xx series. (But that had a totally different,
> SPARC-based controller that was fibre connect to the host. (Though still
> SCSI between the array controller and trays))
>
> --johan
>