Try to answer the following questions, that will help you in setting up your first Elastigroup.
- How long will it take to provision a new server (include time for bootstrapping / data transfer etc)?
- Is your application more memory / CPU intensive or is it fairly balanced?
- What metrics are the most important for the application? For example, do you know how many connections you can have to your server before it becomes unresponsive?
- Do you need scaling policies to meet increased demand?
- What is the optimal server type and size for your application? Keep in mind with Elastigroup you can create a mix of different instance - types and sizes. If your application runs fine on an m3.medium it can also perform just as well if not better on an m4.large.
Applications that are running in Auto Scaling Groups today can easily be imported in order to take advantage of up to 80% cost savings over normal on-demand instances. Once imported, the Elastigroup will function in much the same way as your Auto Scaling Group with the same characteristics.