This Tuesday you may have noticed a fair few websites not really looking… Well. Right. You can thank Amazon AWS S3’s outage in Virginia for that.
For those not in the know S3 is a storage system sold by Amazon which allows you to store stuff, and then make it accessible via a URL. It’s so simple, easy and cheap to use that it’s become widely adopted; sites and apps such as imgur, Twitch.tv, Github, Expedia, Slack, Twilio, Zendesk, Amazon itself, and a huge variety of others all use it.
It also offers cool features, such as the ability to make your stored data available to multiple regions. In fact, it’s encouraged. Unfortunately, due to economic, political or ignorance reasons, you can’t always do this. I can’t point fingers, especially when I know I’ve done such things.
As with any cloud service (and make no mistake, this is the cloud at work) you have to remember that there will be downtime. Cloud providers are not infallible. That being said, depending on your experience and resources, they often still offer greater reliability than you could achieve self-hosting. The trade off is generally far less control, and visibility, when issues occur.
The best advice for businesses utilising cloud services is to expect, accept and plan for failures.
As a minor aside, we find it a little sad that a single company’s outage can affect so many others so substantially, especially considering the web was originally very decentralised. Perhaps the widespread outages we’ve been seeing recently will push things towards that state once again.
And to the teams working on the problem at Amazon – #hugops – we’re sure that it wasn’t fun. Although we are looking forward to reading the postmortem.
Since we’ve written this Amazon have posted the postmortem of the issue, which can be found here.