Clusters and load balancing on AWS (and cloud in general)

EagerDinosaur

I've got a question about how horizontal scaling is achieved in AWS, and maybe on other cloud platforms. By that, I mean how do you add extra servers to increase throughput and availability.

I've got some pre-cloud experience of using clusters of servers to achieve horizontal scaling, using IIS, SQL Server and Weblogic. In all of those situations, the application software was aware that it was operating in a cluster and special configuration settings in the application were needed.

In AWS, it seems the preference is for AWS to take care of clustering, by spinning up new server instances (based on existing AMI images). The individual servers are not aware that they are part of a cluster (or auto-scaling group in AWS terms).

Am I correct in thinking when designing an AWS system, I should generally try to avoid using clustering features built into application servers, and let AWS take care of it via auto-scaling groups?