A network is a system of interconnected things or people. A computer network is a network that consists of computers and their resources, such as shared scanners, printers, routers, and servers. A network load balancer can be one of two things. For one thing, it can refer to a specific Microsoft® technology called Network Load Balancing (NLB) that was part of the family of Windows 2000 Server® and Windows Server® 2003 operating systems and was an optional component for Windows Server® 2008. The term network load balancer can also refer generically to a system that provides for distributing processing evenly across and among available resources.
A cluster is the term for a group of multiple servers that are linked and work in tandem in various ways to accomplish tasks. The cluster can provide a system both with fault tolerance and load balancing. A network load balancer can be employed to distribute load to multiple servers in a network and also to multiple disks in a storage area network (SAN), as well as distribute file protocol requests across file servers. A transaction that enters a network can be divided across all the servers or can be redirected to a server as it becomes available.
One common situation in which a network load balancer is useful is in distributing a large number of incoming HTTP (HyperText Transfer Protocol) requests to the various web servers located in a server farm. This prevents any single server from becoming overloaded. The practice of load balancing helps negate denial of service attacks, the point of which is to overload servers.
A network load balancer can be several different things. It can be a software program, the sole purpose of which is load balancing. It can also be a hardware device. One hardware device that acts as a network load balancer is a multilayer switch, a device that inspects and forwards the packet traffic that enters the network.
DNS (Domain Name System) servers are another type of hardware that can be used as a network load balancer. One technique used is called round robin DNS. In this system, when a DNS request is received, the requests are passed to the servers one after the other in a continuous loop. One issue with this system is that no account is taken of the demands that each transaction will make on the server it is sent to, so it actually acts more like a distributer than an actual balancer, although if there is not system overload, it will have a balancing effect.