In article <100119.141444.6997@avi.netaxs.com> you wrote: : LAX internap cluster was down for about 7 hours last night... Traffic was : still being sent, affected all southern california roadrunner/mediaone, : @home among others... Had reports of down clusters continue to being sent : traffic a LOT of late.. : Matt Traffic should be rerouted in under 20 seconds, and we use an HSRP-like mechanism to snarf traffic if just one machine is down... I've never heard of routing to down regions for longer than seconds. We've seen a fair bit of election-traffic requested being more than end ASs or pieces of end ASs can actually take, no matter where we serve from, but not of serving from down regions. noc@akamai.com can investigate, just let them know the prefixes you're coming from. They'll escalate to the networking group if anything truly strange is going on. Thanks, Avi