How I Got Stung By Amazon EC2

So a few months ago I launched https://socialcrawlytics.com/ and to save a few bucks I thought I should use an Amazon Micro instance to run some workers off, it is free for the first year so it was a good option. Must admit, I really wanted to get on Heroku but Amazon's free tier was much more attractive.<p>For those of you who are interested, Social Crawlytics is based on the following technology:<p>1. Redis 2. FuelPHP 3. Node 4. Pubnub 5. MySQL<p>To cut the chase, within a couple of weeks I discovered that my pet project was attracting a lot of attention and out of nowhere the app was processing hundreds of large reports a day - which meant we needed more workers. So we developed the app pretty much based on Amazon's AMIs, every time there was a huge spike the app would auto launch new instances and instantiate workers until the spike was over and then it would terminate the extra instances.<p>This was obviously an expensive decision because we had instantly moved from a "free tier" to basing pretty much most of the app on Amazon's infrastructure. I knew the costs would add up but I sought solace in setting up Alerts so every time a certain threshold was hit I would get an alert. Turns out that Amazon's alert had a bug and and so did Social Crawlytics, basically in a matter of hours we clocked up over £640 bill and I wasn't alerted by Amazon at all.<p>Luckily, I just logged in to AWS and when I saw the cost I almost had a minor attack. So why did I share this here? Simple! If you are on Amazon watch your account like a hawk, you just never know what the costs would add up to and if you have a bug, it could cost you thousands in matter of hours.<p>I hope this will help someone somewhere.