Dear Akamai, you got a /32 there not a bunch of /48s - how to break Facebook and annoy lots of users

Patrick W. Gilmore patrick at ianai.net
Tue Aug 21 08:04:19 CEST 2012


On Aug 20, 2012, at 16:42 , Roger Jørgensen <roger at jorgensen.no> wrote:

>> As for whether we should deaggregate PA space, I'm afraid that decision is
>> already made.  We are not asking for 1000+ /32s from the RIRs, and there
>> really isn't another good solution to this problem AFAIK.  We are not
>> trying to cause problems, but we have constraints in which we must work as
>> well.
>> 
>> If this does not fit your view of how the world should work, I am afraid
>> we shall have to agree to disagree - unless you can come up with a better
>> solution than asking for 1000+ /32s.
> 
> there is another way to look at this, consider geobased routing... but
> that's like swearing in any church/temple.
> 
> Not to forget it will likely create atleast a dozen other issues we have
> to deal with... like how and who should manage it, and how should it be
> routed etc. Someone will either way have to carry the "covering" prefix
> for any region. But who said moving mountains (or moving the wind) should
> be easy?
> 
> Either way, it is another way to look at this :-)

Unless I am confused (and I may well be - lack of sleep and not completely familiar with the topic), geobased routing does not solve the problem at hand.

Geo-based routing doesn't even begin to allow for things such as "this node is full, move [XX users | YY customers | ZZ $FOO] to node $BAR".  Akamai uses actual performance data to load balance multiple nodes based on latency, loss, cache churn, server type, etc., even including how much the customer is willing to pay vs. how much it costs to serve from each node.

For this we need discrete nodes, with unique addresses, which we control.

-- 
TTFN,
patrick



More information about the ipv6-ops mailing list