[afnog] QoS/TE

[Jim Forster]

>> I've been thinking about this a great deal (and following
>> a similar discussion on the nanog forum).
>>
>> Is QoS/TE in the core really necessary. I mean, like Joel
>> mentioned some time before AfNOG-2005, aside from your
>> edge (customer) and upstream links, what other links does
>> a network expect to really saturate if engineered well?
>
> 	Considering the capacity of most modern core equipment,
> 	I tend to say avoid QoS there.  It's usually cheaper
> 	to upgrade the core than to deploy QoS (including: analysis,
> 	prioritization definitions / tagging, feedback, followup).

I would tend to agree, although most of the thought and analysis has  
been done for places with fiber based cores.   In these places the  
cost of additional bandwidth for core networks is generally cheaper  
than additional bandwidth for access networks.   In Africa it's  
conceivable that the access networks could be cheaper bandwidth  
through various wireless technologies than the cores/backhauls which  
will only slowly move off satellites to terrestial fiber and high  
speed wireless trunks.

QoS is very frequently over-engineered, but if has a place it's on  
the bottleneck links; I'm suggesting that the bottleneck links may be  
in a different segment of the network, depending on the economics.

   -- Jim