[prev in list] [next in list] [prev in thread] [next in thread] 

List:       inet-access
Subject:    RE: fancier ethernet switches
From:       "Pete Templin" <pete.templin () texlink ! com>
Date:       2003-05-19 16:08:00
[Download RAW message or body]

Not bad, until your Riverstone sales person closes your Riverstone TAC tickets because he decided to read your tickets and wants to sell you training instead of letting the TAC fix their bugs.  BTDT.

-----Original Message-----
From: Andy Walden [mailto:andy@tigerteam.net]
Sent: Saturday, May 17, 2003 8:21 PM
To: list@inet-access.net
Subject: Re: fancier ethernet switches



Riverstone ES 500 - Wirespeed, supports OSPF.

andy
--
PGP Key Available at http://www.tigerteam.net/andy/pgp

On Thu, 15 May 2003 jp@pour.midcoast.com wrote:

> I'm curious about people's experience with layer-3 and layer-3/4 ethernet
> switches, as I'm investigating HP's midrange switches like their 5300xl
> series and 2650.
>
> The idea of combining switch and ethernet router in one is attractive, but
> the routing features found on these switches are somewhat limited compared
> to linux or cisco.  As long as the layer-3 switch/routers can do basic
> routing reliably and fast, I could use them at many locations. Many of our
> sites have a linux or cisco router and a managed ethernet switch. It would
> seem that less ethernet "hops" would result from having the switch and
> router as one. I'm not sure any performance advantage would result from
> one fewer ethernet hop for the average broadband customer. Fewer ethernet
> cables are always a plus in terms of reliability.
>
> The place where I have doubts that any of these switch/routers would work
> as switch+router replacements are at our datacenter where we use IP tables
> in linux to produce traffic accounting data (for graphing) for each IP
> address of ours it forwards, or the cisco where we do traffic shaping for
> certain broadband and colocation customers who are recipients of tiered
> bandwidth. Are there any layer-3 switches that can track traffic per IP
> number like Iptables does or can shape traffic per IP address, ip block,
> or per port number (layer-4 switches).
>
> Reliability and cost are obvious concerns - one of the reasons why we like
> our layer-2 HP switches.  (n.b. the exception that proves the rule is the
> hp4000m switches don't like operating reliably in below-freezing
> temperatures - must be something in the modular design expanding and
> contracting in the cold)
>
> We rarely put serious load on an ethernet connection carrying customer
> traffic. Right now, things rarely exceed 10mbps full duplex on a 100mbps
> fd connection. Ping times on the lans are always low. I don't see an
> overwhelming compulsion to try the traffic prioritization and layer4 QOS
> facilities built into these switches. Would they be of any benefit in such
> levels of ethernet traffic utilization? The only benefit I've come up with
> so far would be for controlling a future instance of DOS like the ms-sql
> worm that spread last year. If there are more benefits for QOS on a barely
> loaded ethernet, I'm all ears.
>
> TIA,
> Jason
>
> --
> /*
> Jason Philbrook   |   Midcoast Internet Solutions - Internet Access,
>     KB1IOJ        |  Hosting, and TCP-IP Networks for Midcoast Maine
>  http://f64.nu/   |             http://www.midcoast.com/
> */
> -
> Send 'unsubscribe' in the body to 'list-request@inet-access.net' to leave.
> Eat sushi frequently.   inet@inet-access.net is the human contact address.
>

-
Send 'unsubscribe' in the body to 'list-request@inet-access.net' to leave.
Eat sushi frequently.   inet@inet-access.net is the human contact address.
-
Send 'unsubscribe' in the body to 'list-request@inet-access.net' to leave.
Eat sushi frequently.   inet@inet-access.net is the human contact address.

[prev in list] [next in list] [prev in thread] [next in thread]