netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Ben Greear <greearb@candelatech.com>
To: Karen Shaeffer <shaeffer@neuralscape.com>
Cc: netdev@oss.sgi.com
Subject: Re: cheap 4-port tulip-based cards available
Date: Sat, 19 Oct 2002 14:59:56 -0700	[thread overview]
Message-ID: <3DB1D5DC.6040800@candelatech.com> (raw)
In-Reply-To: 20021019144236.C11659@synapse.neuralscape.com

Karen Shaeffer wrote:
> On Sat, Oct 19, 2002 at 02:32:10PM -0700, Ben Greear wrote:
> 
>>Karen Shaeffer wrote:
>>
>>>On Sat, Oct 19, 2002 at 02:06:51PM -0700, Ben Greear wrote:
>>>
>>
>>I make traffic generators out of them (www.candelatech.com)
>>
>>They are also good for firewalls and routers, as well as more
>>exotic network devices.  Using linux for a switch does not make so
>>much sense these days when you can buy a 16 port switch for under $100,
>>but with bridging code it can be done.
> 
> 
> Actually, I think linux based switches could be quite interesting in
> a pipelined, beowulf-like or openmosix cluster. (Of course, GigE would
> probably be employed rather than FE.) You could then integrate traffic
> shaping/monitoring functionality into the switch that could be designed
> to interoperate with the cluster nodes to assist in optimizing data flow
> and subsequent processing. With openmosix, this could also be used to
> migrate processes towards optimum node placement, based on network
> traffic analysis as well as nodal resource availability.
> 
> It's a subject I've been thinking about quite a bit lately. This is most
> interesting in adaptive, pipelined clusters that will be processing
> real-time streaming data. That's certainly a niche market at the present
> time.

Right now, you'll have a tough time getting a single GigE NIC to run line
speed, full duplex.

However, I've thought about writing a controlling process that monitors
and dynamically configures a switch though SNMP or some other API.  That way,
you could migrate machines across LANs/VLANs to better distribute your traffic.

And, you could keep all your traffic running line speed (or however fast
your switches run)

> 
> cheers,
> Karen


-- 
Ben Greear <greearb@candelatech.com>       <Ben_Greear AT excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear

      reply	other threads:[~2002-10-19 21:59 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2002-10-15 23:52 cheap 4-port tulip-based cards available Jason Lunz
2002-10-16  4:29 ` Richard Gooch
2002-10-19 21:06 ` Ben Greear
2002-10-19 21:14   ` Karen Shaeffer
2002-10-19 21:27     ` Richard Guy Briggs
2002-10-19 21:28       ` Karen Shaeffer
2002-10-19 22:07         ` Richard Gooch
2002-10-19 21:32     ` Ben Greear
2002-10-19 21:42       ` Karen Shaeffer
2002-10-19 21:59         ` Ben Greear [this message]

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:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

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

  git send-email \
    --in-reply-to=3DB1D5DC.6040800@candelatech.com \
    --to=greearb@candelatech.com \
    --cc=netdev@oss.sgi.com \
    --cc=shaeffer@neuralscape.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* 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).