Amazon S3 gets a SLA. Exhale.
A couple of hours ago Jeff Barr (senior evangelist over at Amazon) posted about Amazon S3′s new Service Level Agreement – which if you happen to run services on the platform (like we do here at Webreakstuff) is a pretty good piece of news. Ever since Amazon S3 (or Simple Storage Service) officially launched developers have been asking for an SLA in order to formally guarantee the service’s reliability and Amazon’s commitment to keeping it going.
Some of the developers building applications with Amazon S3 have been asking us about an SLA, or Service Level Agreement. An SLA defines the minimum acceptable level of performance from a service along with some sort of penalty for not meeting expectations. A typical SLA actually defines a performance or reliability boundary which is somewhat lower than what the system is actually designed, built, and expected to deliver.
We know that many of our customers, including a multitude of teams within Amazon, are using S3 in mission-critical ways and need a formal commitment from us in order to make commitments to their own users and customers.
And the agreement looks good, too. Amazon will give you 10% service credit if uptime goes below 99.9% and 25% credit if it goes below 99% in a given month. Which tells you a lot about how reliable they believe their platform really is.
The agreement is in effect since October 1st, which means those of you who’ve wondered (for so long now) whether it would be a safe bet to host something on S3 can finally exhale. Now, and although I do trust Amazon’s reliability – I mean, it’s Amazon -, it’d be great to have an SLA for EC2 as well, but I assume that’ll be up when it officially launches.