Network Ring

Justin Shore justin at
Tue Sep 8 21:06:42 UTC 2009

sthaug at wrote:
>> Rod Beck wrote:
>>> What is EAPS?
>> A joke of a "standard" and something to be avoided at all costs.  I 
>> would echo the last part about Extreme switches too.
> Disagree. I don't believe anybody would claim EAPS is a "standard"
> just because an RFC has been published. 

Pannaway does.  That was one of the very arguments I used against their 
product when they were brought in.  They claimed that it was a standard 
because it had a RFC.  I tried to explain the difference between an 
Information RFC and a Standards Track to no avail.  Of course this also 
came from the Pannaway SE that gave me 3 quotes I repeat as often as 
possible to as many people as possible.  He said:

1) that we didn't need to run an IGP across our network because we 
weren't big enough to need one.  This was in response to my query about 
their lack of support for IS-IS.  He said that he'd seen SP networks 
many times our size get by perfectly well with static routes.

2) that we didn't need QoS on our network if our links weren't 
saturated.  I won't get into the holy war over serialization delay, 
micro bursts, and queuing here.  It's been hashed out many times before 
on NANOG I'm sure.

3) that IPv6 was just a fad and that it would never be implemented in 
the US.  I got our /32 in 2008 and am working on the deployment now. 
I'm certainly not breaking new ground here either.  It may not be the 
most common thing in the US but it is picking up steam for everyone not 
running Pannaway products since they don't support IPv6 (the BASs and 
BARs that we ended up buying at least).

> As for Extreme switches - they have their strengths and weaknesses,
> just like any other product. We use lots of Summit X450/X450a, for
> L2 only, and have been generally reasonably happy with them. If I
> could buy a similarly featured product from Cisco, for a similar
> price, I might well choose Cisco. But at least in our case Cisco
> *doesn't* have a competitive product (case in point: ME3400 - too
> few ports, too few MAC addresses, funky licensing even if you just
> want to do simple QinQ).

I don't have any experience with the ME3400 unfortunately.  A mix of 
vendors isn't a bad thing if you have the knowledge, depth and time to 
keep up with each of them so you can support the device adequately 
(adequate staffing is involved here too).  When one buys a budget switch 
just to save a few bucks they tend to get what they paid for and none of 
what they didn't (training, experience for their staff, printed 
third-party references, reliable online support groups for example).

I'm in a situation right now where a vendor has proposed a basic L2 
switch solution to redundantly connect 2 of our sites.  They come in 
cheaper than the Cisco equivalent (4 4948-10GEs) but we also have 
absolutely no experience with that vendor.  That means interopt testing, 
future finger pointing in the heat of an outage, double training staff, 
inevitable config errors and typos thanks to the differences between the 
vendor we're used to and the one that is being proposed for this one-off 
connection.  The better fool-proof solution costs a bit more and I have 
to convince management not to save a short-term buck which costs of many 
long-term bucks.  Sometimes you really do get what you pay for.


More information about the NANOG mailing list