IETF Logo Mail Archive

Re: linear search of TCP control blocks on demultiplexing of an incom ing IP datagram

Luigi Rizzo <rizzo@icir.org> Wed, 17 April 2002 18:29 UTC

Date: Wed, 17 Apr 2002 11:29:24 -0700
From: Luigi Rizzo <rizzo@icir.org>
To: "Vishnepolsky, Oleg" <ovishnepolsky@doubleclick.net>
Cc: "'tcp-impl@grc.nasa.gov'" <tcp-impl@grc.nasa.gov>
Subject: Re: linear search of TCP control blocks on demultiplexing of an incom ing IP datagram
Message-ID: <20020417112924.A49970@iguana.icir.org>
References: <03E81ABDC8C2D31192C9009027D5B86E08ADB6CD@NYC-EX03>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <03E81ABDC8C2D31192C9009027D5B86E08ADB6CD@NYC-EX03>
User-Agent: Mutt/1.3.23i
Sender: owner-tcp-impl@grc.nasa.gov
Precedence: bulk
Status: RO
Content-Length: 990
Lines: 20

On Wed, Apr 17, 2002 at 02:06:10PM -0400, Vishnepolsky, Oleg wrote:
...
> used linear search except I used a one control block cache lookup since most
> of the TCP packets come in "trains".

well, this seems to be more a feature of your test environment than
a general property. It may be true on a workstation-type of system,
where there are only very few active connections at once (but then
the PCB lookup load is probably negligible compared to the cycles
spent doing DCT and XML parsing and bytecode interpretation), but
certainly not in a server-type of system with hundreds of active
sessions.

	cheers
	luigi
-----------------------------------+-------------------------------------
  Luigi RIZZO, luigi@iet.unipi.it  . Dip. di Ing. dell'Informazione
  http://www.iet.unipi.it/~luigi/  . Universita` di Pisa
  TEL/FAX: +39-050-568.533/522     . via Diotisalvi 2, 56126 PISA (Italy)
  Mobile   +39-347-0373137
-----------------------------------+-------------------------------------

v2.8.7 | Report a Bug | By Email