Load Balancing

The best way to scale a web application is by creating multiple servers and balancing the incoming traffic and load between them.

Traditionally load balancing was very costly to implement, typically requiring expensive hardware load balancers (two of them if you wanted high availability), each with their own rackspace, cooling and power costs. On a hardware model most organisations deployed two or three web servers which spent most of their time idle; if there was a peak surge the delay in commissioning additional hardware meant that the response to increased demand was very slow.

We are able to offer highly flexible and inexpensive load balancing. Since every aspect of the CronoSphere platform is protected by HA, a deployed load balancer is automatically made “highly available”. There is no hardware to buy, no rackspace and power to consider. Webservers can be added in seconds by making copies (clones) of existing servers. This flexibility leads to a highly agile operating environment that only pays for resource when its required.

Unlike some other commercially available balancers, or other cloud providers our load balancers are not subject to “success tax” where you are required to pay more for additional licenses, throughput or connections.

Our enterprise load balancers are able to handle all types of HTTP, HTTPS and TCP traffic, using either Layer4 or Layer7 balancing modes. A variety of balancing algorithms are possible as well as session stickiness (stateful load balancing) to cater for applications which are not balancer aware. Layer4 provides for near 10Gbit wire-speed throughput whilst Layer7 with full traffic inspection can handle in excess of 3Gbit traffic loads on a single balancer.

Utilising advanced health monitoring and content inspection our load balancers will automatically remove servers from a web farm if the servers are taken offline for maintenance or suffer some issue. This functionality provides additional resilience as applications are scaled out.