Load Balancer

A load balancing strategy or policy instructs the load balancer on where to send the next incoming request. There are many load balancing strategies available depending on the specific solution, however a few common ones are listed below: Round Robin: The most simple load balancing method where each server takes a turn to receive a request. Least Number of Connections: The load balancer will keep track of the number of connections a server has and send the next request to the server with the least connections. Note: Older, layer4 only load balancers tend not to support this as they typically run DSR (Direct Server Return) and don’t know how many connections are currently on the backend servers. Weighted: Typically servers are allocated a percentage capability as one server could be twice as powerful as another. Weighted methods are useful if the load balancer does not know the real and actual performance of the server.

Load balancers distribute application traffic based on many different load balancing strategies or load balancing policies as they are sometimes called. In order to understand if a backend server is online and healthy, a load balancer will use back-end server monitoring and Server health checking. The principles of load balancing have been around for many years but these devices have evolved significantly from the basic layer4 device to much more sophisticated layer7 Application Delivery controllers, or ADCs as Gartner refer to them. ADCs offer many additional key features including security and traffic management.

Load Balancers run server health checks against web servers to determine if they are alive, healthy and providing service. Server health monitoring is the key to delivering resilient applications, and depending on the solution chosen, some load balancers are able to use layer7 health checks which offer greater sophistication in their problem detection. Below is a summary of the different methods of server health checks. Ping: This is the most simple method of server health check, however it is not very reliable because the load balancer can report that the server is up, whilst the web service can still be down. TCP Connect: This is a more sophisticated health check method which can check if a service is up and running. An example of this is services on port 80 for web. Simple HTTP GET: This method of server health check will make a HTTP GET request to the web server and typically check for a header response such as a 200 OK. Full HTTP GET: This server health check will make a HTTP GET and check the actual content body for a correct response. This feature is only available on some of the more advanced load balancing solutions but is the superior method for web applications because it will check that the actual application is available. Customisable Server Health Checks: Some load balancing solutions are able to accommodate custom monitors for TCP / IP applications for better control over their specific application services.

A load balancer acts as the “traffic cop” sitting in front of your servers and routing client requests across all servers capable of fulfilling those requests in a manner that maximizes speed and capacity utilization and ensures that no one server is overworked, which could degrade performance. If a single server goes down, the load balancer redirects traffic to the remaining online servers. When a new server is added to the server group, the load balancer automatically starts to send requests to it.