Server Load Balancing

What is load balancing and how can it help you build a fault tolerant, highly available web site?

Load balancing basically consists of distributing user activity effectively across a computer network so that no single server will be overloaded. The key phrase is, "no single server will be overloaded". Distributing user activity is not enough. You want to be sure a single server does not receive more requests than it can handle. Load balancing can determine which of the servers in a cluster is currently the "least worked". Least worked is determined by the load on that server, such as the CPU usage, how many proccesses are running, the number of users and other system variables which are reported by the server. The user is then transparently routed to the least worked server for connection. From the client's point of view, the cluster appears to be a single server that answers their requests. As traffic increases, additional servers can be added to the cluster.

What happens if one of those servers fail? This is no longer a disaster, as load balancing can also provide high availability by automatically detecting the failure of a server and redirecting user activity among the remaining servers within seconds. Once the server is responsive, it can again share in the work load. Your users will be provided with continuous service, never knowing about a server failure should it happen. In the same manner, load balancing will allow a system administrators to take down servers and perform system maintenance without interruption of services.

Shared F5 Load Balancing

Dedicated Cisco Local Director 430

Dedicated F5 Load Balancer BIG-IP LB

Dedicated F5 Load Balancer BIG-IP HA S

Dedicated F5 Load Balancer BIG-IP HA R