BGP Route Table Entries

[Geoff Huston]

At 7/26/01 07:17 AM, Scott Marcus wrote:
>On Wed, 25 July 2001, Bill Darte wrote:
>
> >
> > Phil Smith's Routing Stats show 104 thousand table entries today.
> >
> > How many existed when BGP first made its debut? Are there stats somewhere
> > for that and for the years intervening?
> > Curious about the curve.
>
>
>
>Sure!  Take a look at:
>http://www.employees.org/~tbates/cidr.hist.plot.html
>
>or better:
>
>http://www.telstra.net/ops/bgp/index.html
>
>
>
>To Lee's earlier comment about routing table growth being a worse problem 
>than address depletion, I would note that growth has been pretty flat 
>since the beginning of this year.  The last I corresponded with Geoff 
>Huston, he argued that this was a purely local phenomenon in his own 
>(Telstra) data; however, I question this as the flattening seems to be 
>present in many routing tables (not all).  Phil Smith has argued the same; 
>CAIDA has also done some interesting measurements that tend to confirm the 
>flattening and suggest that it mostly reflects a drop in new small 
>blocks...  A complicated picture.


I posted on Nanog a few days back that there were some interesting recent 
trends visible:

a) a drop in the growth of AS numbers being announced
b) a drop in the growth of /24 prefixes
and
c) a hiatus in the growth of the number of entries in the BGP FIB tables in 
a number of ASs

Most of these have visibly kicked in round late April one way or another

My take on this is that there are two factors at play at present:

a) a 'clean up' of the routing noise in the /23 - /32 prefix area

b) a side effect of a more general economic slowdown


regards,

     Geoff