Suppose we have the following network:
100 100 10 100 100
R1 <===> SW1 <===> SW2 <===> SW3 <===> SW4 <===> R2
All switches are just L2 devices and all connections are just plain ethernet.
As you can see there is a 10 Mbps link between SW2 and SW3, while all other
links are 100 Mbps.
I need to give priority (& maybe reserve bw) to voice traffic between R1 and R2.
Does it make sense to try to use QoS (LLQ/CBWFQ) on the routers only? If yes
(that would be the best for me), how can the routers
"understand" the actual congestion between SW2/SW3, when all intermediate
switches use their own (default ?) buffering/queuing
mechanisms?
Will there be any difference on the "congestion feedback" that routers get, if
instead of 4 switches, i have 8 switches?
On the other hand:
Should i try to implement QoS on the SW2/SW3 switches only?
Or even better, do i need to implement QoS at both routers & switches?
If the SW2/SW3 connection is EoMPLS, does it make any difference? Do i need
some kind of QoS on the underlying MPLS network too?
--
Tassos
_______________________________________________
cisco-nsp mailing list [email protected]
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/
