Network Routing without Cisco or Juniper?

It has been a long time since I have seen this thread hashed out,
so I figured I'd bring it up publicly.

Is anyone comfortable using (in a network with > 5 routers) any
non-Cisco or non-Juniper routers for BGP speaking? (Zebra/Gated
boxes only count if customer traffic is carried through the device).

Historically, some networks have been big fans of Gated boxes and
such because they were cheap and scalable to a point. Now many
BGP speaking platforms are extremely inexpensive (even from CSCO
and JNPR). I was wondering if the consensus has changed.

Boxes like Foundry, Extreme, Redback and many others all talk BGP
(at least to a first approximation) but is their lack of use in
the core/edge/CPE a lack of scale, stability, performance or just
interest?

Thanks,

Deepak

[snip]

Boxes like Foundry, Extreme, Redback and many others all talk BGP
(at least to a first approximation) but is their lack of use in
the core/edge/CPE a lack of scale, stability, performance or just
interest?

One Dutch ISP that shall remain unnamed (and is not one I work for or
have worked for) deployed Extreme on AMS-IX, with Extreme's BGP
implementation.

It broke horribly. The Extreme BGP implementation, instead of sending
their peers just their own prefixes, would send each peer *all*
prefixes and then withdraw all but their own networks. However, doing
this with tens of peers at the same time was too much for the Extreme
itself, which died.

Extreme has supposedly fixed this bug, but this ISP switched to
Juniper for routing.

From what I see around me, Juniper for routing and Extreme for

switching is a popular combination. It seems both are considered to be
good at one thing and bad at the other.

Greetz, Peter

With Extreme, it's certainly (in my experience only) a matter of
horrific stability in their routing engine: great switches, truly
scary as routers. The thought of BGP on Extreme is almost comedic,
considering I'm afraid of their RIP...

Yours,

J.A. Terranson

:: Boxes like Foundry, Extreme, Redback and many others all talk BGP
:: (at least to a first approximation) but is their lack of use in
:: the core/edge/CPE a lack of scale, stability, performance or just
:: interest?
::

Foundry makes a very good, very stable bgp speaker. I've had them in my
network alongside cisco's and juniper's for a couple of years now, and
i've never run into any bgp implementation problems that i would consider
major. A few annoying bugs here and there, but nothing significantly worse
than C or J.

Beyond the fact that not too many people are familiar with foundry's
gear, I tend to think that foundry has lost face in the service provider
world for non-bgp related issues. ACL problems and CAM size issues have
come up in really large installs (multi GBps, hundreds of thousands of
flows, etc). Foundry is also behind cisco and juniper in features - GRE
and netflow/sflow come to mind.

The ACL and CAM issues are supposedly fixed in foundry's jetcore chipset
boxes, but i haven't seen any of those yet. Sflow is now an option, and
from what i hear, their implementation is very very good. Overall, foundry
still makes a good box - when you figure in the cost factor, it becomes a
great box.

I've also played with extreme, but the last i checked, they were *way*
behind foundry/cisco/juniper in terms of their bgp stability and feature
set. Overall my experience with extreme has not been a pleasant one. I
know some people who love them however, so who knows. They seem to make
good/fast layer 2 gear, but i've had some scary results with their layer 3
stuff.

-jba

And another NL ISP - Demon - has used:

PC-based routers running gated. At low traffic volumes, they worked
   very well.

A supplier I don't think I'm at liberty to name. When they were good,
  they were very, very good. But when they were bad they were horrid.

Another supplier I don't wish to name. Mostly worked, but crashed if
  you made even the slighest configuration change.

We're now on Junipers and very happy.

Then again, AMSIX and their Foundry's break every other day as well :slight_smile:
In their case, the while is definately noy more then the sum of their parts :slight_smile:

Paul

The AMS-IX does not do BGP on their Foundry's (afaik l2 only)

Another box I personally feel is very overlooked is Riverstone. They
make an excellent box, the CLI is incredible (especially for maintenance
windows. When will Cisco learn to have a Scratchpad or a commit
feature?), and all-in-all they are a very feature rich box. The only
*major* problem I had to do with BGP actually was a fault of their being
RFC-Compliant. I believe this was about a year ago, they dropped the
peer on a bogus prefix, that was being carried throughout the net
(Originating from a Qwest client if I remember correctly.) Then again, I
believe this affected more vendors than just RS.

Derek

From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On Behalf

Of

jeffrey.arnold
Sent: Wednesday, September 04, 2002 5:31 AM
To: Nanog
Subject: Re: Network Routing without Cisco or Juniper?

:: Boxes like Foundry, Extreme, Redback and many others all talk BGP
:: (at least to a first approximation) but is their lack of use in
:: the core/edge/CPE a lack of scale, stability, performance or just
:: interest?
::

Foundry makes a very good, very stable bgp speaker. I've had them in

my

network alongside cisco's and juniper's for a couple of years now, and
i've never run into any bgp implementation problems that i would

consider

major. A few annoying bugs here and there, but nothing significantly

worse

than C or J.

Beyond the fact that not too many people are familiar with foundry's
gear, I tend to think that foundry has lost face in the service

provider

world for non-bgp related issues. ACL problems and CAM size issues

have

come up in really large installs (multi GBps, hundreds of thousands of
flows, etc). Foundry is also behind cisco and juniper in features -

GRE

and netflow/sflow come to mind.

The ACL and CAM issues are supposedly fixed in foundry's jetcore

chipset

boxes, but i haven't seen any of those yet. Sflow is now an option,

and

from what i hear, their implementation is very very good. Overall,

foundry

still makes a good box - when you figure in the cost factor, it

becomes a

great box.

I've also played with extreme, but the last i checked, they were *way*
behind foundry/cisco/juniper in terms of their bgp stability and

feature

set. Overall my experience with extreme has not been a pleasant one. I
know some people who love them however, so who knows. They seem to

make

good/fast layer 2 gear, but i've had some scary results with their

layer 3

Foundry makes a very good, very stable bgp speaker. I've had them in
my network alongside cisco's and juniper's for a couple of years
now, and i've never run into any bgp implementation problems that i
would consider major. A few annoying bugs here and there, but
nothing significantly worse than C or J.

Thinking of it, I want to confirm, although we have only really used
IBGP (including IMBGP, and doing MD5 authentication) and OSPF on
those (please, no flames that you only need either of those :-).

In this respect the Foundries have never been problematic, and I
noticed they learned the full routing table much faster than our (old)
C's upon startup. The only problem we had was that in our deployment
we really needed MBGP, and that became available much later than
originally announced. But when it came it instantly worked as
advertised, at least as far as we tried.

Beyond the fact that not too many people are familiar with foundry's
gear, I tend to think that foundry has lost face in the service
provider world for non-bgp related issues. ACL problems and CAM size
issues have come up in really large installs (multi GBps, hundreds
of thousands of flows, etc). Foundry is also behind cisco and
juniper in features - GRE and netflow/sflow come to mind.

My main problem is that I find debugging protocol operation (such
as PIM-SM) much more difficult than on Cisco. And you can't expect
them to have as many resources to develop new feeeeatures all the
time; and the ones that get the resources may not be those that are
interesting to ISPs.

The ACL and CAM issues are supposedly fixed in foundry's jetcore
chipset boxes, but i haven't seen any of those yet. Sflow is now an
option, and from what i hear, their implementation is very very
good. Overall, foundry still makes a good box - when you figure in
the cost factor, it becomes a great box.

Definitely agree. Also they start up incredibly fast, because the
software is so small. So upgrading software on the box is relatively
painless.

I'm a big fan of both Foundry and Riverstone, as BGP speaking routers. I've
had great luck with both. Foundry has some annoying bugs at first, but these
seem to have been resolved. I recommend both.

- Daniel Golding

I can't say personally I like the Foundry's. When I was testing (granted
this was 2 years ago.) I saw some traffic when they were under near
wireline load and holding full tables. I'm sure their vastly improved,
just as Riverstone has. But Foundry's CLI just doesn't cut it for me.
Again, the Foundry bias is fairly old, and I have a couple in the lab
waiting for me to test new code. But I can't go back to their CLI after
using RS for so long.

Derek

From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu] On Behalf

Of

Daniel Golding
Sent: Wednesday, September 04, 2002 1:23 PM
To: jeffrey.arnold; Nanog
Subject: RE: Network Routing without Cisco or Juniper?

I'm a big fan of both Foundry and Riverstone, as BGP speaking routers.
I've
had great luck with both. Foundry has some annoying bugs at first, but
these
seem to have been resolved. I recommend both.

- Daniel Golding

>
>
> :: Boxes like Foundry, Extreme, Redback and many others all talk

BGP

> :: (at least to a first approximation) but is their lack of use in
> :: the core/edge/CPE a lack of scale, stability, performance or

just

> :: interest?
> ::
>
> Foundry makes a very good, very stable bgp speaker. I've had them in

my

> network alongside cisco's and juniper's for a couple of years now,

and

> i've never run into any bgp implementation problems that i would
consider
> major. A few annoying bugs here and there, but nothing significantly
worse
> than C or J.
>
> Beyond the fact that not too many people are familiar with foundry's
> gear, I tend to think that foundry has lost face in the service

provider

> world for non-bgp related issues. ACL problems and CAM size issues

have

> come up in really large installs (multi GBps, hundreds of thousands

of

> flows, etc). Foundry is also behind cisco and juniper in features -

GRE

> and netflow/sflow come to mind.
>
> The ACL and CAM issues are supposedly fixed in foundry's jetcore

chipset

> boxes, but i haven't seen any of those yet. Sflow is now an option,

and

> from what i hear, their implementation is very very good. Overall,
foundry
> still makes a good box - when you figure in the cost factor, it

becomes

a
> great box.
>
> I've also played with extreme, but the last i checked, they were

*way*

> behind foundry/cisco/juniper in terms of their bgp stability and

feature

> set. Overall my experience with extreme has not been a pleasant one.

I

> know some people who love them however, so who knows. They seem to

make

> good/fast layer 2 gear, but i've had some scary results with their

layer

I have to second that. Riverstone is definitely a solid box.

Featurewise, routing protocols are excellent, but services are not quite
there. (I.E. it doesn't support any IP tunneling protocol in any shape or
form. GRE is extremely useful under some circumstances, but sadly,
not with riverstone).

I'm currently running Ciscos at the edge, but Extremes in the core for
our datacenter hosting - iBGP only, mind.
The Extremes are a little odd, in that one refuses to talk BGP at all
(!) but other than that I've not had any real problems.

Unless I had a specific reason not to use "known kit" (read: Juniper
or Cisco) at the edge, I'd probably stick with the usual - but that's
more out of familiarity than genuine worries.