Amazon's S3 experiences outage

Update at 8 p.m. PDT: Amazon's Service Health Dashboard reported around 5:25 p.m. PDT that U.S. and European service "has been fully restored" and that the company "will provide more detail on this event once we have completed a full investigation."

Amazon.com's Simple Storage Service--a major component of its online computing services--is apparently experiencing problems Sunday.

The e-tailer's "Service Health Dashboard" reports that the S3 service in both the United States and Europe is experiencing "elevated error rates."

The outage is causing CenterNetworks images to break, the site reports, and SmugMug reports that "a large portion of the photos and videos stored" on S3 are currently offline.

SmugMug seemed sympathetic to Amazon's woes:

Every component SmugMug has ever used, whether it's networking providers, datacenter providers, software, servers, storage, or even people, has let us down at one point or another. It's the name of the game, and our job is to handle these problems and outages as best we can.

The online storage service had its public launch in early 2006. One of the things that makes S3 attractive to start-ups and power users is that it ties in with other Amazon Web Services like the elastic computing cloud and its SimpleDB service. Using all three, start-ups can offload some of the tasks that usually required spending a large amount of money up front to do this work--saving them, and potentially their customers time and money.

[shelbyb]

Why does Amazon.com not have image issues when S3 goes down? I assume they are their own customer.

[caplan]

I believe Amazon uses Akamai for image storage; S3 is not really a CDN in that it does not provide geographical location-sensitive delivery.

[enovikoff]

Amazon has amassed an impressive series of breakdowns in the last year. Together with its terms of service that offer no guarantee of uptime, I'm curious why businesses would trust their mission-critical IT to it. When I talk to my potential customers, they're enamored of Amazon because of it's slightly lower per-unit cost than competing cloud/outsourced computing offerings, but when you factor in all of the downtime as well as other hidden costs such as management, coding around their system's peccadilloes, redundancy to compensate for their lack of persistent images, charging for bidirectional bandwidth, etc. it seems to me that much of Amazon's price is hidden, sort of like the larger part of an iceberg.

Eric Novikoff

[porterme]

While Amazon's service dashboard is nice, I much prefer CloudStatus.com for noting outages. While Amazon has *always* been forthcoming on outages, third-party verification is almost always better than merely relying on self-reporting.

Cheers.
Matthew

[jeremymhogan]

CloudStatus.com reported the outage twenty minutes before Amazon's own system status noted it. Nothing against Amazon, as outages are a fact of life, but users of CloudStatus.com had a full twenty minute head's up on other AWS users.