[c-nsp] Debugging T1 Bounces on CT3

[Pete Templin]

Robert E.Seastrom wrote:

> Sidebar regarding the CAC M13 muxes: Everyone I know who uses 'em
> loves 'em, and they retain pretty high value on the secondary market,
> which suggests that this is not an isolated phenomenon.
> 
> Did you like your WideBank 28s?  I am realistic enough to realize that
> there are good reasons (spares standardization, etc) to swap out
> working hardware that you like in favor of something that is the same
> as the rest of your network...

Don't take this the wrong way, but "that's not my department".  A 
co-worker handles facilities, transport, and voice, while I handle the 
IP/MPLS side of the house (he's got five folks beneath him at four POPs, 
and I've got one grunt beneath me, so the transport side isn't something 
I can safely take over).  For the most part, the WideBank 28s were quite 
reliable, but we did have an assortment of failures here and there.  I 
think the nails in the coffin were the overall challenge of density 
(heat load caused us to go every other RU, lots of DC power connections 
to deal with) and network management (telnet access was horribly slow, 
and keeping track of addresses vs. DS3 was deadly).

The co-worker who handles the transport side has a habit of getting 
"addicted" to the shiniest widget/vendor and pushing to get their stuff, 
without conducting a thorough test or preparing a careful rollout, so 
it's entirely possible that we took a step backwards in selecting the 
MX2820s.  I've been campaigning for months to just get ONS 15454s and do 
our cross-connects there, rather than on DSX1 panels, and apparently we 
may be going that way soon enough.

But admittedly I was a bit surprised at the relatively low net cost to 
roll out the MX2820s and trade in the WB28s...

pt