Gawker: valleywag, ec2

Amazon.com back after over an hour offline

Jackson West — Fri, 06 Jun 2008 15:20:00 EDT

After a considerable outage that started around 10:30 a.m. Pacific Time, Amazon.com's homepage is now back up and still trying to get someone, anyone, to buy a Kindle. Shortly after the outage, the company issued a short statement in the forums for 3rd party sellers titled "Amazon Un-Planned Event" — not such an unusual occurrence, apparently, but rarely of this magnitude:

We are currently investigating an issue that has impacted the availability of the Amazon.com website. Engineers are actively engaged in resolving this issue and we will provide an update once the issue is resolved.

According to our commenters, Amazon's web services such as EC2, S3 and SQS experienced no outages, so the problem didn't take down dozens of startups with it. Shame.

Amazon.com site outage doesn't exactly inspire confidence in world domination plans

Jackson West — Fri, 06 Jun 2008 13:40:00 EDT

It's one thing when Twitter goes limp (again). It's another when e-commerce oligarch Amazon.com can't keep it up for customers. But that's just what readers are reporting, and we've confirmed. I mean, if you're going to rule the Web, it really is best to remain virile atop your throne. Sure it means lost sales for Amazon — and for its network of affiliates — but our question is, are Web startups using the company's "cloud" services such as EC2 also experiencing problems? Update: Even an attempt to access the site via IP address instead of domain name is for naught, ruling out a simple DNS issue.

Amazon.com's grid-computing service goes offline for 90 minutes, saving its profitless customers money

Jordan Golson — Mon, 07 Apr 2008 13:40:00 EDT

A number of servers running Amazon.com's Elastic Compute Cloud service, which provides pay-by-the-hour computation, went offline this morning from 2 to 3:30 a.m. EC2 is one of Amazon's developer services, offering low-cost virtual servers mostly to startups. Dozens of users complained in this thread on Amazon's message board, where an Amazon staffer reported the "notworking team" — a Freudian slip for "networking"? — was on the problem. What were they complaining about? That their websites stopped losing money for 90 minutes?

One tipster wonders "maybe Amazon trying to sell their shitty 'availability zones' idea" by having instances go down. "Availability zones" allow users to spread their virtual servers across geographic areas to increase availability. The genius of the idea: Sure, Amazon's server clusters should be reliable enough not to need it, but now Amazon can sell downtime as a value-added feature.

As for the outage, Amazon reports the problem was with external connectivity and was a networking problem — nothing to do with the servers themselves.

Amazon.com gives startups a 50-percent-off sale

Owen Thomas — Wed, 27 Feb 2008 16:20:32 EST

Jeff Bezos likes to say he's in the business of delighting customers. And then he delivers that howling, hooting laugh. The latest guffaw-provoker: Amazon EC2, a service which lets startups run their programs on servers housed in Amazon.com's datacenters. When it launched, Amazon promised "the equivalent of a 1.7GHz x86 processor" — in other words, a fairly low-powered server, but at the cost of a dime an hour. Ted Dziuba, the acid-tongued former editor of Uncov, found that Amazon actually delivered half that performance. Why haven't you heard more about this? Likely because most of the me-too, slapdash websites making use of Amazon's EC2 aren't running anything more processor-intensive than an index-hit SQL select.

Update: SmugMug's Don McAskill, also an Amazon customer, thinks it's all a big misunderstanding. The short version: Not all gigahertzes are created equal.