On Mon, Apr 01, 2002 at 10:44:48AM -0800, Don Westlight wrote:
> Hey all,
>
> When I joined this list several months ago, I was hoping to tap into
> a vibrant conversation about the use of Foundry hardware in
> meaningful environments: eg campus core, internet core, research
> labs, etc. Anybody interested in participating in such conversations
> is hereby invited to speak up:
>
> Our Summary
> Installation: NWAX.NET, a traditional regional Internet exchange point
> [2] NetIron 800 boxes
> Services: Layer 2, IP Multicast
> Hardware satisfaction: very good
> Support & software satisfaction: jury still out
> Contemplating: another NetIron for layer 3
>
> Questions:
>
> (1) Does anybody out there use Foundry boxes for core routers ¯ eg,
> full internet routes? (eg. anybody from NANOG here?) If so, is this
> a plain vanilla install?
No, though I'm curious how well that works. We used to use Ciscos for
that (7206, 7206 VXR), and now we use Junipers (M5) and we couldn't be
happier than this.
> (2) Does anybody out there do this with advanced services ¯ eg, IP multicast, QOS/MPLS, or anything else complicated and wacky?
...
> (3) Does anybody out there have the sort of relationship with Foundry (some sort of non-disclosure) where the merits of any particular code implementation can be discussed rationally... ? Alternately, is this an issue for you? Is anyone else out there confused about which commands live in which code train (B2S, B2P, etc)
...
> (4) Does anybody use the Foundry proprietary SNMP MIB rather than MIB-2?
> We've got a real question with this by the way ¯ we're hoping to use the proprietary 64 bit counters, but the output in at least one case is in *octets* not something rational, so MRTG barfs ¯ we're in conversation with Foundry about this, but have not heard much yet...
Isn't octets pretty much normative in SNMP for traffic moved over an
interface? We've pretty much migrated off MRTG to Cricket. Not perfect
itself, but nice. Anyway, no answers here, sorry.
> (5) I'm curious to see what percentage of people on this list use
> Foundry boxes for purely layer 2 devices, and what issues turn up in
> this environment as well (for example inability to put access lists
> on the command interface...)
I'll give you a different answer - we use them purely for L4, i.e. we
use the Foundry ServerIrons for creating virtual server clusters.
We did have a bit of a learning curve in terms of having to figure out
what they wouldn't do. However, once you learn the limitations of what
you can and can't do with them, it's pretty much configure them and
forget about them. If you want to set up a server for some IP service
with above 99.999% visible uptime, this is the way to go. (Use the
symmetric priority feature for failover of your virtual server between
two ServerIrons, put multiple real servers behind the primary one, and
you're set.) At that point the main hazard is self-inflicted damage,
e.g. mirroring a bad configuration onto all your servers.
-- Clifton
--
Clifton Royston -- LavaNet Systems Architect -- cliftonr@lava.net
"What do we need to make our world come alive?
What does it take to make us sing?
While we're waiting for the next one to arrive..." - Sisters of Mercy
This archive was generated by hypermail 2b29 : Mon Aug 04 2003 - 04:10:05 EDT