Amazon S3 is a service that many websites use to host videos, pictures and other data. It’s run by Amazon.com.
Until today it had a near spotless reputation for reliability. We are testing Amazon S3 at WorldTV for experimental future features.
Many sites we like such as Twitter and SmugMug, and any number of data backup sites, use Amazon S3 to run a large part of their service. Today at 12.31pm GMT one of three Amazon S3 datacenters went down completely, causing widespread hurt for many of these sites. Just over 2 hours later, Amazon had fixed most of the problem.
The failure is disheartening for many people involved in this business, including myself, because prior to today we had a nice touchy feeling about Amazon and their product. It offers smaller teams the ability to ‘box with the big boys’.
Amazon are not the only one who do what they do, after today other ‘in the cloud’ operators are going to get much more of a look in. To operate ‘in the cloud’ is possible. But it’s the old adage, ‘you can’t put all your eggs in one basket’. You can, but you have to figure your service will be down sometimes and you will be hopeless to do anything about it. Running a second backup with another operator is possible, but this adds complexity and precious time for developers who are making the next generation of web applications.
To their credit, Amazon have been quick to fix the problem, and made several pledges to improve matters for future in posts to the Amazon S3 forum.