On Fri, 14 Nov 2003, Suresh Ramasubramanian wrote:
Stephen J. Wilcox writes on 11/14/2003 7:16 AM:
So anyway, was discussing the cidr report at the last nanog.. I was pointing out that deaggregation is discouraged by the naming and shaming and then someone else pointed out that this list has scarcely altered in months.
So, what can we do as the operator community if this report isnt having the desired effect?
Stop accepting /24 type routes?
Please no... That will drop me off the map..
Yeah maybe but what about where the RIRs have assigned independent /24 space.. or ISPs have subdelegated the IPs to a multihomed customer, was more thinking about where a bunch of routes originating from a single ASN can be aggregated rather than routing bloat in general. There are numerous such examples of people with eg a /19 announcing 32x /24 etc
Steve
I don't have the stats handy at the moment, but we decided to Multi-home I researched several issues with /24 blocks. One thing that seemed to stick out was that some providers were using /20 and /21 as "multi-home" blocks. They were reserving that block just for /24 multi-homing.. and I also remember that of the /24 being annouced independently, a majority of them were not multihomed....... just how bad is the auto-summarization at the upstream for the route propagation via BGP in the large routers anyway? Jim
On 14 Nov 2003, at 14:41, McBurnett, Jim wrote:
just how bad is the auto-summarization at the upstream for the route propagation via BGP in the large routers anyway?
What auto-summarisation? If you're talking about the cisco "auto-summary" command, then the answer is "so bad that it's universally disabled" (but then "auto-summary" is concerned with aggregation on pre-CIDR, classful boundaries which I don't think is what you were talking about). Absent a precise understanding of all the routing policies concerned, proxy aggregation is dangerous and is hence generally not done. Joe
participants (2)
-
Joe Abley
-
McBurnett, Jim