[ppml] Effects of explosive routing table growth on ISP behavior
bjohnson at drtel.com
Wed Oct 31 15:49:02 EDT 2007
Scott Leibrand wrote:
> Brian Johnson wrote:
> > This is true today; I can break-up my ARIN allocation and advertise
> > longer prefixes down to a point where I will lose some network
> > reachability. Actually, we should remove the language from the IPv6
> > policy that indicates anything about advertising the "covering
> > aggregate" unless we are willing to cede that all assignments will
> > remain purely hierarchical in nature. So anyone who multi-homes will
> > have to get an assignment from a RIR to maintain reachability.
> I don't think that's true. Today, anything you advertise in IPv4 down
> to a /24 will be accepted more or less by everyone.
Yes. But this came about due to common network policy and time, not ARIN
policy. It could just as easily been a /20 if the screaming was not as
> However, that's
> only a convenience for optimal TE. In order to just maintain
> reachability, all you really need is for your upstreams (all your
> transit providers and all their transit providers) to accept your
> deaggregated routes from you and from each other.
I have no control beyond my agreement with my direct upstream. If I am
connected with provider A and B for multi-homed up-links, provider C
(connected with both A and B) has no obligation to take anything I say.
If I advertise /24 and they have a block on anything longer than /22,
I'm not reachable from a network connected only by C. This is not an
ARIN issue; it's a management and network policy issue.
> The rest of the world
> need only accept the RIR-assignment-size covering aggregate, route the
> traffic toward one of your transit providers, and then let the
> more-specifics take over when they hand the traffic off to them.
This assumes that you have advertized the actual "RIR-assigned-size"
block to your providers. This is sometimes not desirable and I have seen
it often where the larger block is not advertized.
Since IPv6 was "engineered" to get back to a more strictly hierarchical
model, this forces us to find different ways of handling these issues.
My only real point is that routing decisions should not be directly
"governed" by ARIN policy. I think we are just beating this topic with a
smelly herring. :)
More information about the ARIN-PPML