RE: draft-ietf-multi6-multihoming-requirements-06.txt

[Vijay Gill <vijay@umbc.edu>]

On Thu, 12 Jun 2003, Randy Bush wrote:

> An Ops-Dir reviewer had the appended comments.  excuse the droll
> phrasing.  any chance we could address them?
>
> randy
>
> ---
>
> Let's see: load balancing is not a realistic requirement.  For the
> architecture to be automatic, we have to be dynamic and we've
> proven that we don't know how to do that yet.  Even to give people
> some tools seems to lead to abuse.  Who controls the balancing?
> The source?  The destination?  Don't say 'both'.  No packet can
> serve two masters.


The load balancing does not have to be automagically done. For example,
with the current mechanisms available to us in v4 + BGP, a large promising
local dialup provider used to make extensive use of selective
announcements, local preference, selective prepends etc to ensure that the
traffic to its various transit providers was evenly balanced since most of
the billing was done on the higher on in+out and so effectively there was
the opportunity to optimize cost by moving someone's higher outbound to
another provider who had a high inbound.

> Performance.  Exact same arguments since the only way to affect
> performance is to load balance.


I think this has been addressed in the satellite link for big bandwidth
latency insensitive transactions (ftp, nntp) but terrestial links for
interactive traffic example.

> Policy: so ill stated as to be meaningless.  If this is a call
> for policy routing, that's fine, but it has nothing to do with
> multihoming.

see example above of cost optimizing using mechanisms available in the
underlying protocols to move traffic around according to the policy
"minimize cost"

I think abley has covered the rest.

/vijay