[c-nsp] XENPAK packet loss

[FF]

Ok. On the 4-6 ps/nm/km basis we are close but not outside out budget.
The lower number is what  we budgeted for.

We solved the primary traffic problem (which packet loss on two
completely different links). It wasn't related to the hardware
but rather the MPLS FIB being in exception. Supposedly you can't clear
this without a reboot, but clear mpls ldp neigh * seems
to do the trick.

Exactly *why* so many routes were being tagged as MPLS isn't clear yet.

Cisco TAC wasn't more lucid than saying bad things can happen in MPLS
exception state, and you have to reboot to fix it.

Thanks for your assistance everyone! I will be doing some dispersion
measurements on the other leg.


On Tue, Jun 16, 2009 at 11:50 AM, Marian Ďurkovič<md at bts.sk> wrote:
> On Tue, 16 Jun 2009 16:19:08 +0200 (CEST), Mikael Abrahamsson wrote
>> On Tue, 16 Jun 2009, FF wrote:
>>
>> > I thought Chromatic Dispersion is distance related. This is supposed
>>
>> Yes it is.
>>
>> > to be SMF-28 DSF, the optics are supposed to be 80km (XENPAK DWDM 1600
>> > ps dispersion tolerance). Do you need a DCU even when operating within
>> > that range? One of the links is only about 40-50km.
>>
>> Well, you didn't say the links were DC or not, and you didn't say how
>> long the link was. At 40-50km, CD is most likely not the cause of your
>> problems.
>>
>> According to some info I found, SMF-28 DSF still has 17.1ps/nm/km,
>> meaning your 1600ps dispersion tolerance only gets you 94km? I might
>> be wrong though, I can't get the whole article, google only displays
>> from its cache.
>
> Wait, if it's really DSF, it has zero dispersion at 1550 nm.
> But such fiber is unsuitable for DWDM operation.
>
> If it's NZDSF, it should have ~ 4 - 6 ps/nm/km.
>
>    M.
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>



-- 
FF