namelesstavern.net

google.com down~

google.com back up~

Turns out it wasn't down for everyone ><.

for google.com to be totally down would mean the internet is more or less totally down. They are probably the most redundant website on the internet at the moment.

Google had some issues a few weeks ago that were a bit more widespread. I was thinking this was something intentional more then a hiccup.

Depends what route you take. I have two gateways here, one through Qwest which was problematic and one through XO which was fine.

They had some massive outages again today. Several locations across the US were "down" until route changes completed to push folks to japan, uk, etc.

I think both XO and Quest just resell lines owned by VB or ATT. Google is claiming the issue today was route fails on ATT part.

Both XO and Quest are major internet backbones, any routing issues on either would be their own hardware and not ATT, though its certianly possible that ATT actually owns the fiber itself. Most of them major backbone carriers as well as some of the major datacenters in the US have direct peering with google, the only downside to this is that if they have an issue, anyone routing through them to google will have issues until the routes are corrected.

Having had to work to deal with ATT routing fails myself (specifically at that point with what was originally CERFnet down in San Diego that ATT had just bought a few months before that), I can readily believe ATT having routing snafus.

That said, do you know how hard it is to find someone at ATT who not only knows BGP but has access to major routers in order to correct issues with it, even as a major corporate customer with a dedicated contact? Ugh.

-Arlos

ATT has very few people in their internal networking team who really understands BGP, thankfully of our 230Gb worth of backbone links, not a single one is directly through ATT.

Well it wasn't the telco's after all, google pony'd it up and said they screwed up. Not sure how the claim of only effecting 14% can be messured. It sure appeared to effect alot more than that.

Even if it took out one of their entire datacenters, the others were still functioning. They probably just based the 14% on 14% of their front end servers were behind the area where they had the issue.