Horizontally Scaling a web based application

Question

I have a web application which runs 3 services simultaneously: "A", "B" and "C" and they're sharing the same data from an external database.

"A" may be using more resources than "B" and "C", and might need its own dedicated machine.

I'm thinking to scale the application horizontally (by deploying new machines, instead of updating the current servers' configuration) in the following way:

• measure the load on the server @ "every minute" intervals
• if the load is over 90% for more than 30 minutes, deploy a new machine running a new instance of service "A", "B" or "C", depending which has the highest load.
• if the load of the currently monitored machine is lower than 10% and it's not the only machine running, shut down the machine.

Is there any book or recommended website on this topic as well?

How about software tools to help? (Or more specific, I might consider Amazon EC2)

Many thanks, Vlad

Answer 1

Amazon's Elastic Load Balancer seems exactly what you need. It will allow you to scale your instances if they are running in EC2. It suports HTTP/HTTPS but other non-web TCP services.

RightScale would also be an option. It also supports other clouds in addition to EC2.