Elastic Load Balancing is designed to provide an easy-to-use fault tolerant cloud platform. Using Elastic Load Balancing, you can automatically balance incoming traffic, ensuring that requests are sent to an instance that has the capacity to serve them. It allows you to add and remove instances as needed, without interrupting the operation of your cloud. If an instance fails or is removed, the load balancer stops routing traffic to that instance. If that instance is restored, or an instance is added, the load balancer automatically resumes or starts routing traffic to that instance.
If your cloud is serving traffic over HTTP, the Elastic Load Balancing service can provide session stickiness, allowing you to bind specific virtual instances to your load balancers.
Elastic Load Balancing can balance requests within a cluster, or across multiple clusters. If there are no more healthy instances in a cluster, the load balancer will route traffic to other clusters, until healthy instances are restored to that cluster.