Re: patent question

[Rod Van Meter <rdv@cips.nokia.COM>]

It's worth looking up the patent at www.patents.ibm.com.  Steve 
Blumenau might even be lurking on this list; comments, Steve?

My reading of this (and certainly I'm no lawyer) is that it's only 
about having two computers that already have SCSI devices attached to 
them pass SCSI commands between them to be proxied to the attached 
devices.  It has nothing to do with having the devices themselves speak 
SCSI over a network protocol. It does appear to intend to apply to any 
network protocol, not simply IP. Mentions ethernet a lot.

So, as I see it, it's possible that this could affect host 
implementations, but doesn't seem at all relevant to device-side 
implementations of iSCSI. Seems to me that it's pretty obvious once you 
have a device that speaks SCSI over a network that the host has to be 
able to package up the commands to send to it. I suspect it's only 
EMC's remote mirroring that's truly defended by this patent.

The patent doesn't really have a "related work" section, but I find it 
unlikely that Blumenau was unaware of things like SCSI-GPP and 
VAXclusters' MSCP, which have been discussed here before, even if he 
wasn't aware of the research at LLNL or ISI.

The submission date on the patent, btw, is Jan. 14, 1998, and it was 
issued Nov. 30, 1999.

			--Rod


Greg Finn and Cheng-Ann Tan wrote:
> 
> Patents are often issued that are in error.  Does it fall afoul of
> prior art?  What is the date on the patent?
> 
> IP/SCSI was prototyped and published pre-IETF years ago.  Talk to
> Rodney Van Meter regarding dates.
> 
> If the patent is more general than just IP, then other link-layer
> protocol work encoding SCSI would equally apply as prior art.  For
> those dates talk to others on this list.

> hi, all:

> i did a patent search on the net and came across emc's
> patent (5996024) which describes using scsi devices
> over network protocols. it covers ip, of course.
> 
> pardon my ignorance, but if a patent has already been 
> granted, how would this affect further work on i-scsi,
> ip storage, etc?
> 
> -cheng