First login?
Lost password?
 
 
sctp-discussion AT lists.franken.de 
SCTP Discussions

Text archives Help


Re: [SCTP-Discussion] Bundling


Chronological Thread 
  • From: "Brian F. G. Bidulock" <bidulock AT openss7.org>
  • To: Prince A P <princearimpur AT huawei.com>
  • Cc: sctp-discussion AT sctp.de
  • Subject: Re: [SCTP-Discussion] Bundling
  • Date: Thu, 22 Jul 2010 02:58:38 -0600
  • Dsn-notification-to: <bidulock AT openss7.org>
  • List-archive: <http://lists.franken.de/pipermail/sctp-discussion>
  • List-id: SCTP Discussions <sctp-discussion.sctp.de>
  • Organization: http://www.openss7.org/
Prince,

On Mon, 19 Jul 2010, Prince A P wrote:

> In the case of bundling enabled(for new data sending), may be
> for a packet of 20 chunks, the receiver will send one SACK.
> Since the Cwnd increase is based on SACK count, the increment
> is slow in comparison to non-bundling case.

That statement is untrue: cwnd increase is not based on SACK
count. See RFC 4960 Section 7.2.2. A SACK does not cause an
increase in cwnd unless it acknowledges cwnd or more bytes and
there were cwnd or more bytes outstanding before the SACK
arrived.

Therefore, assuming that 20 chunks fill an MTU:

1) No bundling: the sender sends 80 packets, 1 chunk per packet,
one packet at a time. The 80 packets must be sent before the
first SACK arrives; also for each arriving SACK another one
or two packets must be sent, 1 chunk each, (otherwise cwnd
will not increase at all). After the 80'th SACK arrives
(assuming for argument's sake that each packet is SACK'ed),
cwnd will increase by 1 MTU.

2) Bundling: the sender sends 4 packets, 20 chunks per packet,
one packet at a time. The 4 packets must be sent before the
first SACK arrives; also for each arriving SACK another one
or two packets must be sent, 20 chunks each, (otherwise cwnd
will not increase at all). After the 4'th SACK arrives
(again assuming no delayed SACK), cwnd will increase by 1
MTU.

In the case of a bandwidth-limited link, bundling will increase
cwnd faster because only data is counted for cwnd calculations
and not packet headers. No bundling wastes the avialable
bandwidth with an excessive number of packet headers.

--brian

--
Brian F. G. Bidulock ¦ The reasonable man adapts himself to the ¦
bidulock AT openss7.org
¦ world; the unreasonable one persists in ¦
http://www.openss7.org/ ¦ trying to adapt the world to himself. ¦
¦ Therefore all progress depends on the ¦
¦ unreasonable man. -- George Bernard Shaw ¦


Archive powered by MHonArc 2.6.19+.

Top of page