RE: network controls are necessary

["Tony Hain" <alh-ietf@tndh.net>]

Joel M. Halpern wrote:
> ...
> In one regard, you end up needing significantly more logic in 
> the host to make any kind of intelligent choice.

By 'intelligent' you mean 'the same policy choice as the network admin',
right? This is not really a technical problem as much as a domain of
control problem. The address selection rules allow for local
administration. So if a network manager can convince the local host
manager to set source selection policy to prefer a specific order from
the list of prefixes in an RA, "the intelligent thing will happen". If
the network admin can't do that, the hosts are effectively uncoordinated
traffic sources that any transit ISP has to deal with already.
Attempting to dictate operational policy through a standards process is
going to fail.
  
> This means more code, more state, and 
> more things for the network administrator to configure 
> (probably incorrectly.) Secondly, you couple the hosts into 
> the behavior of the rest of the routing 
> system, reducing the ability of the system to cope with 
> changes (either 
> network changes or protocol changes.)

This looks like a big problem to the network admin, because they are
used to dealing with a few boxes, but for a host admin, this is exactly
what they do. What needs to happen here is for them to work together
(*gasp*).

> 
> If we really want the hosts to make the choice (a concept I 
> am doubtful of) 
> I suppose we could invent a query / response protocol for the 
> purpose of 
> asking a routing intelligent server what source /dest pair 
> from a given set 
> of sources and dests would be a good pair to use.

If you want the ability to change policy in real-time from the network,
it makes much more sense to put options in the RA than create a
query/response protocol. 

> 
> I strongly prefer the notion that other entities in the 
> system would make 
> the intelligent choice about the source address to use, and 
> could change 
> that choice as necessary.

Even though accomplishing that will require changing all the application
software globally???

Tony


> 
> Yours,
> Joel M. Halpern
> 
> At 08:46 AM 12/6/2002 +0100, Erik Nordmark wrote:
> >I did not suggest that hosts should run BGP. That would be 
> completely 
> >silly. But IPv6 hosts are supposed to have a source address 
> selection 
> >table according to a draft in the IPv6 WG (soon RFC). If the exit 
> >routing policy can be expressed with a few rules it would 
> essentially 
> >be additional rules in that table (and a protocol by which the hosts 
> >can learn those rules). Hence my questions on the list (so 
> far without 
> >answer) about reasonable sizes for the exit router selection 
> >policy/routes.
> >
> > > a) The hosts to have the same routing capabilities as routers 
> > > currently have.
> >
> >Sorry I don't follow the logic. The hosts don't route. Hosts with 
> >multiple source locators that talk to a node with multiple 
> destination 
> >locators just do a selection e.g. when creating a new connection.
> 
> 
>