RE: iFCP as an IP Storage Work Item

["ELLIOTT,STEPHEN (HP-Roseville,ex1)" <stephen_elliott@hp.com>]

Regarding the statement of JP Raghavendra Rao:

It would be nice if somebody comes up with a stronger case for connecting an
iSCSI host to a FC device - Is this attempted for the survival of FC or for
a speedier deployment of iSCSI ?

I offer the following technical consideration.  Consider extremely large
storage arrays.  I think storage arrays tend to be FC devices.  Moreover, 
in some cases there are limitations on the number of SCSI devices that 
can be accomodated on a single host, which would prevent software RAID 
with SCSI from scaling to the same degree as FC in these cases.  

There are other devices as well that are FC-based in order to integrate 
within the FC-SAN that in some cases includes these extremely large storage
arrays.

In each of these cases, connectivity to these devices is essential.  
What SAN devices do you envision ISCSI connecting to?  In what timeframe
do you think the FC SAN market will disappear?

It seems that this issue is separate from the iFCP/iSCSI debate, 
because an appropriate iSCSI bridge could accomodate these 
SCSI-over-FC devices. 

Regards,

Stephen Elliott

-----Original Message-----
From: JP Raghavendra Rao [mailto:jp.raghavendra@india.sun.com]
Sent: Thursday, January 04, 2001 8:19 AM
To: ips@ece.cmu.edu
Subject: RE: iFCP as an IP Storage Work Item



> I believe iFCP should be an IPS work item for the following 
> technical reasons:
> 
> 1)  iFCP allows leverage of existing FCP-based driver stacks and
> preservation of the $$$ and @#$!!% that have been invested in them
> by vendor companies and their customers.
> 

I think the argument for preserving software doesn't make strong case in the
face of a migration to a new mapping/tunneling protocol, new software and
new
administration challenges in spite of the fact that all of this is likely to
mimic FC and contained in one or two edge routers - Today's FC network is
difficult to administer and any bridging technology to a different
interconnect
is only going to compound it.

It would be nice if somebody comes up with a stronger case for connecting an
iSCSI host to a FC device - Is this attempted for the survival of FC or for
a speedier deployment of iSCSI ?

-JP