Looking to Cut Costs?

Who isn’t? More often than not, there is money to be saved from right-sizing your infrastructure. At stack.io, we recommend performing this type of initiative on a quarterly basis and at least semi-annually.

So What Are Some Common Mistakes?

We can all fall foul of some simple fixes that just get overlooked. From our experience, here are the top mistakes when it comes to optimizing your costs:

• Not rotating your logs - this can pollute your system and potentially eat all of your disk space

• Not updating the OS on your VM instances

• Provisioning instance types that are larger than required

• Not taking advantage of autoscaling - this in turn leads to the over-provisioning of infrastructure resources to handle the peaks

• The lack of comprehensive monitoring tools - while tools provided by the OS can help with troubleshooting ongoing issues, they are not best equipped for looking at historical data.

• Short retention periods - to find the right balance between performance and cost optimization, you need to retain a certain amount of metrics. (We’d suggest at least a week)

What can Stack.io do for you?

Cost optimization doesn’t just happen overnight - it typically requires a few stages to complete this project thoroughly.

Studying Cluster Metrics

We typically look at application servers, non-managed DBs, non-managed load balancers, workers, request/limit settings, autoscaler settings, and more. We’re looking for:

• Graphs displaying the resource usage from instances

• Logs on the servers

• Outputs from utility CLI tools on the server

• Configuration files or network configurations to make sure there are no security issues (such as the DB being open to the internet)

• Infrastructure setup - to see if it is set up to be highly available or not