A week in the life of Amazon EC2 Server

Photo credit: http://www.freeimages.com/photo/city-scape-1230395

If you are a server but with nothing to serve, then life can be pretty idle and your availability status should be Green, right? But wrong if you are an EC2 server, because Amazon is quietly putting other Virtual Machines on you, to keep their costs down. This hardware sharing is how a public cloud works and can offer services at a cheaper cost.

All is well until it is not, i.e., when other users on your server, without your knowledge, start to do some high bandwidth activities such as reading or writing a large file. To check this we ran Meghafind scouts on an Amazon server for 5 days, doing nothing else but simply monitoring its performance once/minute. On an average things looked Green as shown below.

However, if we look at the maximum scout scores, indicating highest level of activity on an hourly basis, the picture became dramatically different as seen below. It means that EC2 server was busy serving other users, so if we really had a website or another job running, it will crawl.

The status of a Server’s performance follows traffic light pattern, i.e., turning from Green to Yellow, Orange and finally Red as the load increases and persists for longer periods of time. This is determined by the timed sequence of Scout measurements showing a server slowdown. A slower machine means loss of revenue or unhappy customers.

Meghafind can alert, and helps you to decide which machine to use. It also offers detailed alert tables with data from the CPU, Memory and Storage loadings on the Server being monitored. This can be a gold mine to debug what went wrong and when.

Stay tuned for how to use Scouts for predicting your application’s performance on a new server.