SAM definition of SCSI Device

Bob Snively bob.snively at EBay.Sun.COM
Wed May 26 16:13:50 PDT 1999


* From the T10 Reflector (t10 at symbios.com), posted by:
* Bob Snively <bob.snively at EBay.Sun.COM>
*

> >
> >Actually, for those who are truly dedicated to "dynamic multi-pathing"
> >a la ESCON, Fibre Channel provides the feature through its hunt group
> >definitions with no change to SCSI or to FCP-2. 
> >
> To emulate the capabilities we are talking about here, both the initiator
> and target must belong to (model 3 or 4) hunt groups. Since no switch I can
> purchase supports hunt groups, this may not be meaningful...
> 
> Further, hunt group models 3/4 require that the exchange (and for 4,
> sequence) management is embedded in a common shared entity. This would
> appear to require an HBA and a controller (rather than switch) to manage
> the hunt groups. This perhaps would not require SCSI or FCP-2 changes, but
> would require architectural changes to the controllers and hosts.

Precisely my point.  Full dynamic pathing capability a la ESCON requires
precisely the specified functionality either from the FC hunt group, or
worse yet, from all the independent but associated SCSI ports of a host and/or
peripheral device.  There is no HA, switch, or peripheral with anything
resembling such a capability, and little reason to build one given the
architectural properties of the SCSI and Fibre Channel interface.
That is why I remain firmly opposed to any structure beyond the conceptually
simple, but architecturally very capable, SCSI architecture we now have.

Bob

*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at symbios.com





More information about the T10 mailing list