[nsp] ip load-sharing per-packet - cef accelerated ?

[Greg Steele]

> On the other hand, we have 2600s which are unusable with MLPPP (due to
> the CPU hit of multiple E1s), but handle the load just fine with per-
> packet load balancing. YMMV.
> 
We have had interesting experiences showing that 2651's take a terrible
CPU hit on per-packet but 2691's do not. This is on ATM PVC's.

...Greg

 -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -  -
Gregory (Greg) E. Steele              
Senior Network Engineer               (Internet):steele at oar.net
OARnet  
2455 North Star Road                  Voice: 614-728-8100 x203
Columbus, OHio  43221                 FAX:   614-728-8110