From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Sat Nov 13 2004 - 11:15:07 GMT-3
At 4:27 AM -0500 11/13/04, Scott Morris wrote:
>The load balancing done with multipath is still handled via your
>switching/caching method. Which means it's more of a flow-based balancing.
>So when one flow chooses one path, it will go via that path the whole time.
>
>I don't believe the docs state per-packet load balancing at all... That
>would be bizzarre.
<Jack Nicholson mode>
You don't want bizarre?
Then you can't handle the CCIE lab!
</Jack Nicholson>
Seriously, per-packet, as opposed to per-flow, goes beyond bizarre
into operational insanity. The Internet-Drafts I've mentioned are
only a start on the idea in NANOG and IETF discussions.
This archive was generated by hypermail 2.1.4 : Thu Dec 02 2004 - 06:57:43 GMT-3