[afnog] Weekly Routing Table Report

Mark Tinka mark.tinka at seacom.mu
Mon May 6 19:02:24 UTC 2013

On Monday, May 06, 2013 07:11:28 PM Patrick Okui wrote:

> The question is what can one do? It's not like an RIR can
> refuse to give you address space because you have
> de-aggregated your current address space.

Routing was never part of the RIR's tasks, and while they do 
understand and appreciate the issue of RIB/FIB bloat, it's 
technically beyond their scope/mandate.

> Many people
> running workshops on the continent are preaching the
> aggregation gospel but that's as much as I'm aware of.

I suspect, that in Africa any way, there could be a co-
relation between the need to multihome and the ability to 
acquire symmetric bandwidth from several providers of 
relative global connectivity. In the past, de-aggregation 
due to load balancing has been largely linked to secondary 
links not being of the same size as the primary one.

Naturally, de-aggregation will always stay, especially as 
redundancy goes into networks and operators are forced to 
break address space up to keep services running independent 
of geography. But if folk can get access to decent 
connectivity where they can get, say 100Mbps from provider A 
and 100Mbps from provider B, I think there could be an 
incentive to aggregate for such a scenario.

> As for the reasons for deaggregation:
> 	- not much was done in IPv6 to make multihoming 
> 	  effective.[*]

It really is IPv4-world with bigger address space.

LISP, SHIM6, ILNP and friends are all interesting ideas at 
the very least, but operators have real networks to run for 
which they - at least in the short-to-medium term - mostly 
require parity with IPv4.

> 	- RPKI takeup is slow but at least is an option 
> address hijacking.

See DNSSEC - but this can be reduced much more today, if 
operators insisted on only accepting routes their customers 
were actually allocated.

Sadly, this isn't a uniquely-African problem.

> 	- I haven't heard anything that spares you from 
> block blacklisting (based on your announcements) due to
> spam sourced from some clients in those blocks.

The Chinese have had their address space blocked by several 
operators over the years for different reasons. 

Not unexpected in a community that (AFAIK) is not centrally 
governed :-).

> Heh. As do lies, damn lies, and statistics :-)

I think the vendors are also now starting to enjoy the 
memory bloat issue a lot less than they used to as well.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://afnog.org/pipermail/afnog/attachments/20130506/c23e8372/attachment.sig>

More information about the afnog mailing list