Re: FCP over IP item on the agenda

[julian_satran@il.ibm.com]

Jon,

I am taking the liberty of forwarding this to the mailing list.
In fact I DO NOT WANT FCP to e discussed AT ALL during the limited time we
have.
But the new WG chairmenn  David Black (EMC) and Steve Belovin (AT&T)
suggested we do on their draft agenda and that generated my answer.
I still think it is a very bad idea.

Julo

Julian Satran IBM Research Laboratory in Haifa

"Jon William Toigo" <jtoigo@IntNet.net> on 16/07/2000 22:28:03

Please respond to "Jon William Toigo" <jtoigo@IntNet.net>

To:   Julian Satran/Haifa/IBM@IBMIL
cc:
Subject:  Re: FCP over IP item on the agenda




Mr. Satran,

I have been reading in earnest the traffic regarding iSCSI and storage over
IP.  I cannot understand why you would want to obfuscate the important work
you are doing to include a debate or discussion of FCP over IP, as is
proposed by EMC (see below).  Clearly, a few companies have already
deployed
FC-based storage networks, but have discovered the great expense, learning
curve, and difficulty in building, scaling and managing the architecture.
That, in a nutshell, is why IP SANs are so interesting.

FC-based SANs confront numerous problems that I have addressed in articles
for the trade press.  (See SANs and Sensibility, InfoStor Magazine, May or
June issue and my current column on storage at PlanetIT.com.)  To bridge
"island FC SANs" across the corporate Ethernet backbone, an assortment of
gateways and bridges will shortly come to market -- including a joint
Brocade/Cisco solution that will likely involve a blade in a Cisco router,
and standalone products from NuSpeed, NetConvergence and others.  IPS
should
focus on IP SANs initially, then embrace the needs of those who want to
bridge their storage solutions to IPS-defined SANs.

In my view, for what it is worth, the handwriting about FC is already on
the
wall.  It is not a true SAN technology, if only because FC is a fabric and
not a network per se.   Efforts to do SANs using FC have run afoul of the
technology's lack of in-band management, lack of interoperability between
vended products, and large staff retraining requirements.  In the end, FC
may provide a nice technology for use within array cabinets, but it does
not
provide a sound basis (in its current manifestation, at least) for an
enterprise SAN.

For what its worth.

Jon William Toigo, consultant and author
The Holy Grail of Data Storage Management
jtoigo@intnet.net

----- Original Message -----
From: <julian_satran@il.ibm.com>
To: <Black_David@emc.com>
Cc: <ips@ece.cmu.edu>
Sent: Sunday, July 16, 2000 4:48 AM
Subject: FCP over IP item on the agenda


>
>
> David,
>
> I've seen the proposed agenda of the BOF on the IETF site.
> I think that we have far less time than we asked for but we may try to
make
> up for this.
>
> I was also under the impression that we are going to discuss thing that
> where
> already published on the IETF and /or IPS board.
>
> Considering the limited time we have - and the fact that we went through
> the FCP
> over IP thing several times in public forums and designer group I wonder
> why
> it appears as a tentative item on your agenda.
>
> If we have to consider new facts we have to have time to digest them and
we
> did not see the yet on any forum (unlike all other items).
>
> I feel very uneasy about having to give up precious time during the only
> face-to-face
> session we have to an item that has yet to be brought to us.
> I think this is also contrary to the IETF practice - or the little I've
got
> to know about it.
>
> Again I have nothing against discussing again FCP over IP - but in order
to
> see if there
> are arguments for it that we did not consider before I think we need time
> off line
> and getting it a first hearing during the only session we have will do a
> disservice
> to the community.
>
> Regards,
> Julo
>
>