archive mirror
 help / color / mirror / Atom feed
From: Werner Almesberger <>
To: Alan Cox <>
Cc: David griego <>,,,
	Linux Kernel Mailing List <>
Subject: Re: Alan Shih: "TCP IP Offloading Interface"
Date: Tue, 15 Jul 2003 02:58:13 -0300	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <>; from on Mon, Jul 14, 2003 at 09:05:53PM +0100

Alan Cox wrote:
> load balancer. If you want to argue about using gate arrays and hardware
> to accelerate IP routing, balancing and firewall filter cams then you
> might get somewhere - but they dont need to talk TCP.

One thing that sounds right about TOE is that per-packet overhead
is becoming an issue, too. At 10 Gbps, the critters come flying in
at almost 1 MHz if you're using standard MTU sizes.

On the other hand, replicating the entire infrastructure on some
non-Linux hardware has several problems, even if we don't consider

 - where is the configuration interface ? In the kernel or in
   user space ? What about existing interfaces ?
 - you'll never get exactly the same semantics. Just identifying
   the differences is a very painful process. And again, what
   about existing interfaces ?
 - testing has just become a lot harder

What I think would be more promising is to investigate in the
direction of NUMA-style architectures, where some CPUs are closer
to NICs and whatever data source/sink those TCP streams go to.

Licensing issues, the classical reason for using independent
stacks, can be elegantly avoided on Linux.

Another area are network processors. They could help with fancy
things like Dave's flow cache, but also with fine-grained timing
needed for traffic control. One problem there is that they're
locked away behind walls of NDAs and proprietary development
environments, so one couldn't even begin to properly support them
in Linux. (What can be done is to treat NP+software as a black
box, but I wouldn't consider this a satisfying choice.)

- Werner

 / Werner Almesberger, Buenos Aires, Argentina /

  parent reply	other threads:[~2003-07-15  5:43 UTC|newest]

Thread overview: 25+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-07-14 19:43 Alan Shih: "TCP IP Offloading Interface" David griego
2003-07-14 20:03 ` Jeff Garzik
2003-07-14 20:23   ` Alan Cox
2003-07-14 20:05 ` Alan Cox
2003-07-14 20:30   ` Shawn
2003-07-15  5:58   ` Werner Almesberger [this message]
     [not found] <>
2003-07-15 11:18 ` Alan Cox
  -- strict thread matches above, loose matches on Subject: below --
2003-07-14 21:51 David griego
2003-07-14 20:29 David griego
2003-07-14 20:23 David griego
2003-07-14 20:19 David griego
2003-07-14 20:31 ` Alan Shih
2003-07-14 20:34 ` Alan Cox
2003-07-14 21:53   ` Deepak Saxena
2003-07-17 22:31     ` Bill Davidsen
2003-07-14 19:14 David griego
2003-07-14 19:26 ` Jeff Garzik
2003-07-15 12:42   ` Jesse Pollard
2003-07-14 19:46 ` Alan Cox
2003-07-14 18:46 David griego
2003-07-14 19:02 ` Jeff Garzik
2003-07-14 21:22   ` Deepak Saxena
2003-07-14 21:45     ` Jeff Garzik
2003-07-15  5:27     ` Werner Almesberger
2003-07-14 19:42 ` Alan Cox

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \ \ \ \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).