How to save money on AWS?




Saving money is one of the motivations in moving to AWS (Amazon Web Services), so you ask what else is needed? Your monthly bills can grow pretty quickly, if you are not careful with the server and services you pick on Amazon’s EC2.

One of the first choices you make is the size of a Virtual Machine (VM), and bigger is not always better. You have to right size the VM to match your task and any local server that you tested it on. Next come services such as Storage, Networking, Load balancing etc. which we will address in future. Focus first on picking the right VM for your jobs.

Pick too small a VM, and your job will run out of cpu or memory,  thrashing the server, causing headaches for other users, and above all, resulting in a slow job run. This may cause a loss of revenue or customers for you. On the other hand, pick a larger VM than you need, cause extra expenditure and job may not necessarily run faster.

Let me explain the last point a bit more. We use Meghafind Scouts to measure our local server, and ensure that the one being picked on AWS is at least as good, in term of CPU, Memory and Storage scores. Then we rest easy and let our job run well in the cloud. However, in some what-if experiments when we picked a bigger EC2 server, job often became slower as presumably other Amazon customers started to use it. The bigger the server, more up or downswings were experienced overtime, sometimes up to +/- 30% difference in performance. Other large jobs generally don’t tend to land on smaller machines, so swings are fewer.

Hence, first step in saving money on AWS is to pick the minimal server that you need, to get your job done predictably. This can be accomplished by matching Meghafind Scout scores on your local test system with EC2 server. To paraphrase, if the shoe fits, wear it.

Install and try Meghafind scouts today: