Re: The Cidr Report

Dear Tony,

you wrote:
>
 [...]
> 2) It seems folks don't use it as much as they can as I see very little 
>  positive changes anymore.
>  
>  However, having said this I would like to understand from folks (reply 
>  directly) if you think I should continue to generate this.
>  
 [...]
>  However, it is
>  worth having a quick look for config errors and would encourage providers 
>  to do this periodically.
>  
Well, I do include the mailing lists on purpose in my answer...

I very much support that you continue to generate the CIDR reports, and 
distribute them. The reasons are manifold, and go far beyond sentimental 
feelings about one of the institutional reports on the Internet.

Why did we start this in the first place? To put up a flag against 
exponential routing table growth. And it worked - growth was down to roughly 
linear for quite some time.

However, newer analysis e.g. by Philip Smith or recently by the new RIS 
project at the RIPE NCC shows that the Internet has again embarked on 
exponential growth of the routing table (e.g. I still doubt that several 
*thousand* longer prefixes announced from one /8 make a lot of sense, or the 
tremendous increase of longer prefixes than /24).

Today, the worries are most likely not that much CPU and memory problems of 
the routers any more, but convergence and stability of BGP tables. Craig 
Labovitz, Abha Ahuja, et al. show results from their work which are not 
exactly reassuring...

Anybody may improve the situation e.g. by proper filtering (very efficient as 
proven e.g. by Randy Bush), or correct route flap damping (yes, RIPE-210 
needs a brush up badly - it is on our action list). As important as this is: 
it only helps on symptoms generated by others on the Internet: too many 
unnecessary routes.

I am convinced that it is ever more important today to have a close look at 
the status of the Internet routing table, and where it may be improved - and 
your CIDR report is one keystone in it.

Thanks a lot for generating these reports during the past years, and please 
continue to generate them. A review of the prerequisites as you outlined in 
the first part of your mail does not change anything in the basic statement 
of the report.

Thanks
   Joachim
   (chairman RIPE Routing WG)
--- JS395-RIPE -- standard disclaimer ---