[c-nsp] How does IOS do loadshaing with ECMP?

[Baek, Steven A (US SSA)]

Would the route processor have stateful tracking to ensure the packet
coming across the wire takes the same route back?

If there were two paths to a remote destination, would the rp know which
path was available and which one isn't?


Steve

-----Original Message-----
From: cisco-nsp-bounces at puck.nether.net
[mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Sukumar
Subburayan
Sent: Monday, November 20, 2006 9:00 AM
To: Phil Bedard
Cc: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] How does IOS do loadshaing with ECMP?

Sup720 supports load-balancing using src_ip,dst_ip+universal_router_id
and 
this is the default.

In this particular case below, you have:

mls ip cef load-sharing full simple

'mls ip cef load-sharing full' lets you use 'src_ip,dst_ip+l4 ports' 
instead of the univeral_router_id

However, the addition of 'simple' keyword, allows hardware to use 
the same number of adjacencies as in the IOS CEF adjacency.

Without the 'simple' keyword the HW installs additional adjacency
entries 
to avoid platform polarization.

sukumar



On Mon, 20 Nov 2006, Phil Bedard wrote:

> That command tells the box to use L4 port information in making load
> balancing
> decisions.   I think if you just set that back to the default it will
> only use
> the source/destination hash like you are used to.
>
> Phil
>
>
> On Nov 20, 2006, at 9:45 AM, Joe Shen wrote:
>
>> mls ip cef load-sharing full simple
>
>
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/