Re: Changes to PI Policy?

[Gert Doering <]

Hi,

On Thu, Apr 17, 2003 at 05:04:07PM +0100, Peter Gradwell wrote:
[..]
> rest), or some type of route caching (so you don't have to 
> know about the whole route).

Route caching and backbone providers (read: traffic to very diverse
destinations) do not get along well.

Route caching and network scans (read: Code Red and friends) are a deseaster
waiting to happen.

This is one of the reasons Cisco went away from netflow caching and to
CEF - no caches, deterministic lookup time for each destination.

[..]
> anyway... what I'm saying is that I don't think we can 
> solve the issue here & now in this WG and I don't think the 
> solution is for RIPE to outlaw PI space because it's 
> inconvenient. I think the solution is to make everything 
> Provider Independent and solve the large scale routeing table problem

There are approaches to this, but I personally really dislike them.

It works like the telephone network.  

You have to register your routes at some central entity.

Once per day, you download all available routes to some kind of 
"preprocessing server".  This server will take your topology and peerings
into account, and condense the table into the minimum necessary to 
route everything into the proper direction (remember that most routers
only have a couple of interfaces, and "most stuff" goes out of one
of them).  Then this routing table is downloaded into the router.

It will scale to about unlimited routes, but the cost will be flexibility
and fast failover in case of outages.  To cope with this, networks need
to get much more reliability at L1 and L2 layers.  Just like telephone
networks...

[..]
> Obviously, if nothing smaller than a /24 will not be routed 
> then there is little point in allocating it. (If some one 
> applies for a /24 PI space because they got turned down on 
> a PA request, we have to wonder why the PA request was denied).

Usually because they have only 5 machines, but wanted a /24 "because it's
much nicer, and we're so important".

Gert Doering
        -- NetMaster
-- 
Total number of prefixes smaller than registry allocations:  58512  (58485)

SpaceNet AG                 Mail: netmaster@localhost
Joseph-Dollinger-Bogen 14   Tel : +49-89-32356-0
80807 Muenchen              Fax : +49-89-32356-299