Re: draft-csapuntz-tcpmsgbnd-00.txt

[Bill Main <wmain@gis.net>]

As a general rule, I would hope that we can keep all the "control" bits in the headers
(versus "last" byte or trailers) to make it simpler for the Silicon to process it.

    -Bill Main

Matt Wakeley wrote:

> I was toying around with an idea similar to this.
>
> My idea is, have the first byte in the iSCSI header be "reserved".
> Whenever an iSCSI message (n) is sent, this first byte of the "next" message (n+1)
> would also be sent, marked urgent.  The result is that the TCP segment(s) for the
> message (n) will have the urgent pointer pointing to the reserved byte of the next
> message (n+1), indicating the beginning of the next message.
>
> Or each iSCSI message could be sent with the "last" byte of the message marked
> urgent, having the same result.  The urgent pointer would then point to the last
> byte of each iSCSI message, thus indicating where in the byte stream the next iSCSI
> message header begins.
>
> Comments?
>
> -Matt Wakeley
> Agilent Technologies
>
> Internet-Drafts@ietf.org wrote:
>
> > A New Internet-Draft is available from the on-line Internet-Drafts directories.
> >
> >         Title           : TCP Message Boundary Option
> >         Author(s)       : C. Sapuntzakis
> >         Filename        : draft-csapuntz-tcpmsgbnd-00.txt
> >         Pages           : 4
> >         Date            : 30-Aug-00
> >
> > TCP does not have a mechanism for specifying message boundaries in
> > a stream. This I-D describes a new TCP option and a new way of
> > using the TCP urgent field to specify message boundaries in the
> > stream.
> >
> > A URL for this Internet-Draft is:
> > http://www.ietf.org/internet-drafts/draft-csapuntz-tcpmsgbnd-00.txt