Clarifications of Digital's SCSI Disk Array signals

Edward A. Gardner 522-2247 21-Jan-1994 1356 gardner at ssag.enet.dec.com
Fri Jan 21 13:47:37 PST 1994


First, this is really a SPI topic, not an SFF topic, so please 
address all further replies to the SCSI reflector.

The mail headers were rather cryptic, but I think the original
comments were from a Mr. Treadway of DPT.  To clarify what he
said:

Digital's StorageWorks (tm) storage arrays use four additional
signals to report the condition of a rack of drives (we call this
a "shelf").  These signals are:

Shelf_OK: Indicates that environmental sensors (power,
temperature, etc.) report satisfactory conditions.

Swap: Asserted while a drive is actually being inserted or
removed.  Generated by using varied pin lengths, guaranteed to be
asserted (during insertion) before any SCSI signals make contact
and to remain asserted until after all SCSI signals are seated,
the reverse on removal.

Fault_Clock, Fault_Data: A serial communication path between the
array controller and some logic in the drive carrier.  Allows the
controller to turn on individual drive fault lamps even when the
drive is totally broken.  We use this to indicate which drive is
broken and needs to be replaced.


These are primarily implemented as backplane signals (but then so
is the SCSI bus).  We sometimes use some excess grounds to bring
them out of the array in what is otherwise a SCSI cable.  Whether
the cable contains these signals or grounds (as specified by
SCSI) is an option that varies with different applications and
customers.  I think we only bring them out with the 68-pin cable
and connector, but I'm not sure.

We have given specifications of these signals to a number of
customers.  If there's interest, I'm pretty sure I could obtain a
specification for distribution in the X3T10 mailing.  Within
Digital, we've discussed proposing these signals to the
committee, and concluded (perhaps wrongly) that the committee
wouldn't be interested.

Finally, I'd like to suggest that all signals should remain
"unsued".  Given the net worth of most bus signals, I doubt that
even a lawyer would be interested in suing one.  If anyone
wishes, I'll be happy to offer a motion at the next X3T10 plenary
that all signals should forever remain "unsued".

Ed Gardner
Digital Equipment Corp.




More information about the T10 mailing list