AWS SLA Discussion

Amazon’s AWS service level agreement has been called into question by Arthur Bergman over at O’Reilly Radar.

if your company is based on AWS: What does your disaster recovery plan look like? How do you react if Amazon goes down or if Amazon decides to shut down AWS? What happens next December when a busy holiday season makes Amazon divert bandwidth from S3 to their main business, Amazon.com?

I thought this was interesting given my experience with a problem that caused my sites to fail. It’s good to have my websites up and running all the time so I can eek out a few pennies of advertising revenue now and then. However, it’s not essential and no real harm happens when they go down, except to waste my time when I have to recover.

I now regularly back up my image and databases from my live server and send it all over to s3. I suppose I could also develop even more scripts to download from s3 to another backup service. And given my limited server instances it would not be a problem for me, but if I had a lot of servers running it would take a while to replace them all.

This issue though needs to be put into context of what we’re really trying to manage and that it risk of business interruption. In that context, AWS is likely to be one of the smaller problems compared to another risks that I find in major companies everywhere and that is significant dependencies on a small number of individuals or even one individual who knows how thing work.

If you really want to manage risk associated with using AWS the first question I would focus on is, who knows how AWS works for us and how many of them are there? If you have a very small team who manages your setup at AWS, you probably have a much bigger risk of them getting a better job somewhere else or getting hit by a bus than by Amazon deciding to leave the business.

While coming up with a disaster recovery plan for AWS is important, it may be the smallest risk the DRP needs to address.