Layer 7 Load Balancing

Layer 7 Load Balancing

The most popular load balancers or application delivery controllers mix traditional simple Layer 4 load balancing with the more advanced application sensitive Layer 7 (Taken from the OSI model) content switching technology. With KEMP Technologies this combination has resulted in the award winning LoadMaster load balancers that are available as both hardware appliances and virtual platforms. Key to this success has been:

  • Extremely scalable solutions
  • Highly optimized application delivery network devices
  • A smart combination of the best of both layers 4 and 7 operational practices

How Layer 7 Load Balancing works

Load balancing balances application requests across a number of servers, these servers are often in a pool known as a cluster and the load balancer presents them to the outside world as a virtual server, it accepts requests as the unique interface to the pool and directs the request to the most appropriate server based on algorithms that have been defined by the IT department. The key thing is that the servers are used need to contain the same content.

The most popular Layer 4 load balancing techniques are:

  • round-robin
  • weighted round-robin
  • least connections
  • weighted least connections

Layer 7 switching directs its requests at the application layer, this type of switching is also known as:

  • Request switching
  • Application switching
  • Content based routing

Like Layer 4 load balancing the load balancer presents the server cluster as a single virtual server and distributes the requests based on the performance health check the network load balancer uses to assess the performance of each application. 

Where Layer 7 load balancing differs from Layer 4 is that it is the staff who are responsible for the applications themselves need to ensure that their applications are perfectly tuned for optimal application performance.

Layer 7 load balancing differs from Layer 4 load balancing in a fundamental way because the servers do not to replicate the same content, but effectively “pass the parcel” this allows for fine tuning , here is an example:

  • Server 1 supplies images and graphics
  • Server 2 delivers the content to the site visitor using scripting and content like CSS and HTML
  • Server 3 allows the user to buy the content
  • Server 4 delivers the purchased content