[c-nsp] Per packet Load balancing

[Rodney Dunn]

You shouldn't do it that way.

You can configure CEF to do per-packet load
balancing.

Turn on CEF globally and then on the inbound
interface feeding the downstream equal cost paths
do "ip load-sharing per-packet".

Don't leave it like you have it now.  You are process
switching all the traffic.

Rodney

On Tue, Sep 07, 2004 at 09:06:51PM +0530, Amol Sapkal wrote:
> Hi Guys,
> I am still following the active post of 'BGP Balacned' - thanks for it.
> 
> Now, I have implemented per packet load balancing, by disabling cef
> and route-cache on both the interfaces from R1. Initially too, I had
> disabled ip route-cache from the subinterfaces, but it seems the ip
> route-cache command actually comes in action only after an application
> on a physical interface (and not a sub-interface)
> 
> 
> Though this has achieved the desired result, I am not sure how good it
> is to disable cef and cache on a very busy router.
> 
> Thanks to all for the work arounds.
> 
> 
> 
> On Tue, 7 Sep 2004 17:01:34 +0200, Carlson Per <per.carlson at banetele.com> wrote:
> > 
> > > > What do you see? Is there some loadbalancing or none at
> > > > all? What does 'ip cef <ip-address>' where <ip-adress>
> > > > is the /28-net show? An alternative, but not the recommended 
> > > > one (see posting from earlier in this list), is
> > > > making CEF load-balance per-packet instead of per-destination.
> > >
> > > I do see traffic via R1-R2-CE. But it is less as compared to R1-CE. I
> > > am not exactly aware of the difference in per-packet CEF and
> > > per-destination CEF. Would be grateful for your inputs.
> > 
> > There is a topic named 'BGP Balanced', started August 24th, that
> > discusses different loadbalancing techniques. Rodney Dunn
> > (a Cisco guy) posted an excellent explanation the 25th.
> > 
> > Per
> > 
> 
> 
> 
> -- 
> Warm Regds,
> 
> Amol Sapkal
> 
> --------------------------------------------------------------------
> An eye for an eye makes the whole world blind 
> - Mahatma Gandhi
> --------------------------------------------------------------------