SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    RE: Question



    If we assumme that we are transferring scsi packets over the local LAN where 
    there is no congestion, can we actually send the packets only on IP ???? I 
    have read thru the some doccuments. But i havent got a clear picture. can 
    you also suggest me how to go about in this ?????
    
    Thanks
    Shesha Bhushan S
    Research Assistant
    Arizona State University
    
    
    
    >From: "Sukanta Ganguly" <sganguly@opulentsystems.com>
    >Reply-To: sganguly@opulentsystems.com
    >To: somesh_gupta@silverbacksystems.com,   "shesha bhushan" 
    ><bhushan_vadulas@hotmail.com>, "IPS" <ips@ece.cmu.edu>
    >Subject: RE: Question
    >Date: Thu, 25 Oct 2001 07:21:06 -0700
    >
    >Well TCP's congestion control will still be functional but the the iSCSI 
    >cannot make much use of it as it sits at a higher layer than TCP. What 
    >really is required is a congestion control within the iSCSI layer and run 
    >it over plain IP, that would be the most efficient way of doing it. I agree 
    >to that some amount of TCP like congestion control will be built into iSCSI 
    >but will be much less than the overheads of the TCP layer (their are 
    >multiple ways of achieving reliability and congestion avoidance on UDP), 
    >and also will make useful sense for the iSCSI layer.
    >   I don't know whether it is too late of not, cause I see many companies 
    >spending a huge amount of money doing TCP offloads. According to them iSCSI 
    >is going to benefit out of it. Yeah it make the un-optimal way of doing 
    >SCSI over IP a little more fast by having a dedicated processor, but is it 
    >the most optimal way of doing the work ????
    >   I have nothing against TCP offload. but lets have a good iSCSI solution 
    >rather than patching up a unoptimal one.
    >
    >SG
    >
    >
    >*********** REPLY SEPARATOR  ***********
    >
    >On 10/24/2001 at 2:25 PM Somesh Gupta wrote:
    >
    > >The question is not crazy at all (the answer
    > >might be :-)). New protocols are required (and
    > >really should for everyone's benefit) implement
    > >some congestion avoidance mechanism.
    > >
    > >Currently the easiest way to achieve this is to
    > >use TCP. Unfortunately we do not depend enough
    > >on it, leading to significant complexity
    > >in the protocol.
    > >
    > >Somesh
    > >
    > >> -----Original Message-----
    > >> From: owner-ips@ece.cmu.edu [mailto:owner-ips@ece.cmu.edu]On Behalf Of
    > >> shesha bhushan
    > >> Sent: Wednesday, October 24, 2001 6:39 PM
    > >> To: ips@ece.cmu.edu
    > >> Subject: Question
    > >>
    > >>
    > >> Hi,
    > >>   I am doing a research project at Arizona State University on SCSI and
    > >> iSCSI. I was going through both the protocols. I have a crazy question.
    > >> iSCSI = SCSI/(TCP/IP). But SCSI is a reliable protocol so is TCP.
    > >> so can't
    > >> we just try to send SCSI packets on IP or SCSI/UDP. Any & All
    > >> comments are
    > >> welcome.
    > >>
    > >> Thanks
    > >> Shesha Bhushan S
    > >> Research Assistant
    > >> Arizona State University
    > >>
    > >>
    > >> _________________________________________________________________
    > >> Get your FREE download of MSN Explorer at
    > >http://explorer.msn.com/intl.asp
    > >>
    > >>
    >
    >
    >
    
    
    _________________________________________________________________
    Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp
    
    

    • Follow-Ups:


Home

Last updated: Thu Oct 25 17:17:32 2001
7399 messages in chronological order