so interesting looking at Microsoft and Akamai both went nuts this AM��http://monitor.akl-ix.nz/cacti/graph_view.php?action=tree&tree_id=1&leaf_id=9

But....��http://www.nzix.net/ape.html seems to be just chugging along with no increase. the conspiracy theorist in me says they cant be real as both MS and akamai are on ape and there surely would have been a spike.

so to test theory someone do an IPerf over the IX for a few hours please..

��

On Wed, Oct 14, 2015 at 9:25 AM, Tim Hoffman <tim@hoffman.net.nz> wrote:
Generally we prefer to use bi-lats with any major peers, then just let BGP do it's thing for RSs...

I've only struck one IX in the world that bounces their route servers every time they add or remove a peer though. APE is rather unique :)

On Tue, Oct 13, 2015 at 3:22 PM, Dave Mill <dave@mill.net.nz> wrote:

On Wed, Oct 14, 2015 at 11:17 AM, Nathan Ward <nznog@daork.net> wrote:


Microsoft���s BGP session on APE bounced 3 hours ago, they���ll be preferring the oldest route to your networks, assuming they prefer them otherwise equally.




Looks like Nathan's worked out part of the puzzle.

What do international providers do on our peering exchanges from an AS path point of view? If all exchanges were treated equally then AKL-IX learnt routes would generally be preferred as the route servers don't insert their AS in to the path. Do providers add 1 ASN to the path on routes they learn over AKL-IX to compensate? Of course I'm only talking about route server routes here - BiLats wouldn't be affected.

Cheers
Dave


_______________________________________________
NZNOG mailing list
NZNOG@list.waikato.ac.nz
http://list.waikato.ac.nz/mailman/listinfo/nznog



_______________________________________________
NZNOG mailing list
NZNOG@list.waikato.ac.nz
http://list.waikato.ac.nz/mailman/listinfo/nznog