My thanks to all for their helpful suggestions.
This is a great list! I'll start doing my homework
on CEF; I had wanted to begin using it anyway! :)
Regards,
Mark
- Mark C. Persiko, persiko@bvsd.k12.co.us
- MIS Dept, Boulder Valley Public Schools
> -----Original Message-----
> From: Bruce R. Babcock [mailto:bbabcock@cisco.com]
> Sent: Thursday, October 05, 2000 11:48 AM
> To: Mark Persiko
> Subject: Re: [nsp] EIGRP Load Balancing
>
>
> Have you tried using CEF? It gives much better distribution
> of load (for IP) than Fast-switching, while still maintaining
> packet sequence. EIGRP with variance will also work in
> conjunction with CEF.
>
> Any I'd suggest using PQ-CBWFQ or LLQ rather than carving off
> a separate interface for voice.
>
> Regards,
> Bruce
>
>
> At 04:52 PM 10/4/2000 -0600, you wrote:
> >Our network has five locations each connected via dual leased-
> >lines back to a central location. The first leased-line in each
> >pair is a full T1 and the second leased-line is a T1 with channels
> >split out for voice, leaving 640K for data.
> >
> >We are running EIGRP in AppleTalk and IP as our WAN routing
> >protocol. EIGRP is supposed to load balance automatically
> >between equal-cost paths, at least in IP. For each pair of T1's
> >above, we are treating both members of a pair as equal cost so
> >that both routes show up in the routing table.
> >
> >Reading in Cisco's BSCN classbook about EIGRP, I see that
> >EIGRP load balances IP packets on a per-destination basis when
> >fast-switching is running on an interface. Watching this on network
> >management shows that one line is full while the other is
> little-used.
> >This happens for minutes at a time. I assume I'd get better link
> >utilization by turning off fast-switching on each line, but
> that would
> >drive CPU utilization through the roof on our routers.
> >
> >Does anyone have any ideas on how to achieve better link utilization
> >in EIGRP? Any parameters that I can set? I've tried using
> "bandwidth"
> >but that makes one link disappear from the IP routing table.
> "variance"
> >would make up for that but I would imagine it would leave me in the
> >same position as if I didn't use either command.
> >
> >Thanks,
> > Mark
> >
> >- Mark C. Persiko, persiko@bvsd.k12.co.us
> >- MIS Dept, Boulder Valley Public Schools
>
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:18 EDT