Another S3 outage
Amazon’s Simple Storage Service experienced a fairly hefty outage earlier this year in February and it looks like they’re up to it again as I write this post. As seen on the previous link, the S3 service has some great rates. It also happens to have an SLA which some people might be taking advantage of this time around.
The interesting thing is how I discovered this on a Sunday afternoon while doing some routine web browsing. Turns out I was shopping on a fairly big retailer’s website (officemax.com) only to find that the pages weren’t loading due to something hanging on a third party interaction with sb.sellpoint.com. Hmmm:
rwoodrum@slard:~$ host sb.sellpoint.net sb.sellpoint.net CNAME sb.sellpoint.net.s3.amazonaws.com sb.sellpoint.net.s3.amazonaws.com CNAME s3-directional-w.amazonaws.com s3-directional-w.amazonaws.com CNAME s3-2-w.amazonaws.com s3-2-w.amazonaws.com A 207.171.183.117
Interesting. So I went to my old job’s website www.avvo.com which, like many other web 2.0 companies, find the S3 service a good compromise to a complex media distribution mechanism. Having worked there and all, I also happened to know that they’re using S3 for media storage. No secret there. Sure enough, no images load causing the site to suffer on both performance and aesthetic fronts. Lame, Amazon, lame.
Somehow or another I noticed some time ago that Woot! was also using S3 for hosting their product images. That doesn’t appear to be the case anymore, however. Probably a wise move on their part.
According to the external AWS Service Health Dashboard, it looks like this problem has been going on for over 4 hours. That’s a pretty freakin’ long time. The only service I know which has outages longer than that in their production systems is aspone.com; a garbage third party exchange provider. Now, in no way shape or form am I comparing S3 (or any other AWS service for that matter) to a managed Exchange provider - that’d be a grave injustice - AWS is orders of magnitude more reliable.
I’ve heard from quite a few people previously employed by Amazon that their infrastructure reliability leaves much to be desired. There must be some grain of truth to all that insider knowledge and another S3 outage in the span of a few months is a pretty reliable indicator. More reliable than the service, anyway.
So sorry, OfficeMax, Amazon has prevented me from continuing to shop for tables on your site. 
Update:
Official downtime? 09:05 PDT - 17:12 PDT - 8 hours 7 minutes