While going into the Autoscaling Environment setup in the Amazon AWS , you should keep a note of the following things in order to build a better dynamic environment
- Always tag the instances while creating the Autoscaling group so that all the instances which would come up from the autoscaling group have the tags associated with them so you can easily find them while looking on the console.
- If you are using the session persistence you can handle the session either from the AWS ELB or from the server. If you select AWS ELB , ELB would manage your session on the basis of the time duration specified. If you consider using the server session than roundrobin algorithm used by the ELB won't work effectively. To overcome this problem save the session in the database and using a memcache or some other caching for the session. This will prevent the overloading of any particular instance.
- If you are using some configuration management tools which builds up your instance and deploys your stack and code as the new server comes up than you can take the advantages of the hooks provided by the AWS. This comes under the autoscaling lifecycle. There are both the pre and post hooks i.e. you can introduce a predefined delay before the instance gets attached to the ELB and start serving traffic. Similarly you can setup a delay in which case instance would get de-register from the ELB so traffic won't come on them before actual termination. This is particularly helpful if you want to take out something from the server before termination.
