[c-nsp] 802.1w vs EoMPLS failover time

[Walter Keen]

Sorry, yes.  There is a jitter buffer however only configurable between 
3 and 29 ms.  When we tested it a 29ms, we noted a severe failure of all 
modem and most fax calls through this box.

Phil Bedard wrote:
> Is there a jitter buffer on the RAD boxes you can adjust?  Generally 
> plain voice can deal with a decent amount of latency. If you can do a 
> 50ms or higher jitter buffer, FRR may allow you to not underrun.
>
> Phil
>
>
> On Oct 30, 2009, at 10:55 AM, Walter Keen wrote:
>
>> Sorry, our current situation is that during a spanning tree 
>> switchover, it encounters a buffer underrun error on the RAD box, and 
>> we are looking to see if perhaps a mpls TE tunnel with explicit paths 
>> (2 explicit paths plus a dynamic path) would help matters any as 
>> opposed to just layer 2 vlans.  I'll look into FRR.
>>
>> Phil Bedard wrote:
>>> The part where you said what the RSTP convergence time was got lost 
>>> somewhere.  Just using a tunnel primary/secondary paths may not be 
>>> quicker than RSTP.  If you use FRR protection as well it may result 
>>> in less traffic loss than RSTP.   Some vendors have different 
>>> behavior when the failure is on the actual ingress node than a 
>>> transit node, so you may want to investigate that if you are using FRR.
>>>
>>> Phil
>>>
>>>
>>> On Oct 29, 2009, at 7:09 PM, Walter Keen wrote:
>>>
>>>>
>>>> I've got a jitter-sensitive application (voice DS3 over some RAD 
>>>> equipment) that we are testing, and I've got a rapid spanning tree 
>>>> ring through the below network.  We have it down to during a 
>>>> spanning tree switchover (tested by adjusting the rapid-pvst cost 
>>>> on the trunk interface), and curious if people feel if EoMPLS with 
>>>> a mpls-TE tunnel would provide faster convergence in case of a 
>>>> failure, given a fairly vanilla OSPF as the IGP, and 2 explicit 
>>>> paths defined (A-D, then A-B-D), as the endpoints of this 
>>>> application are at A and D.
>>>>
>>>> I think I'm going to start testing this tomorrow or next week, but 
>>>> curious if anyone had any thoughts or suggestions.  HW is 
>>>> 7600/RSP720 at A and B, 7600/SUP720 at D and C, all with 6724sfp 
>>>> cards for core-facing interfaces, and 6148 card (10/100) for 
>>>> RAD-facing interfaces.
>>>>
>>>> Network looks like
>>>>
>>>> A-------------------D
>>>> \------B-----------/
>>>> \----------C-----/
>>>>
>>>> Or, A has a connection to D, A has a connection to B and C, B has a 
>>>> connection to D, C has a connection to D.
>>>>
>>>>
>>>> _______________________________________________
>>>> 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/
>>>
>>
>> -- 
>>
>>
>> Walter Keen
>> Network Technician
>> Rainier Connect
>> (o) 360-832-4024
>> (c) 253-302-0194
>>
>

-- 


Walter Keen
Network Technician
Rainier Connect
(o) 360-832-4024
(c) 253-302-0194