So NAT allows a single device, such as a router or server, to act as an agent between a public network such as the internet and a private network. The owners of the private network usually will not want to expose all the IP addresses of the internal servers to the public so what happens is that one address is exposed to the public and this one address is then mapped onto multiple internal servers who can handle the public requests. This in turn means that only a single unique IP address is required to represent multiple servers. 

A example is where a website has one IP address but web page requests are load balanced across multiple web servers who respond to the web page requests. Each of the web servers will have their own IP address but these private addresses will not be visible to the browsers making the web page requests. The browser only sees the one public IP address. This gives the browser the impression that there is only one computer or server handling the request but in reality there can be tens or even hundreds of web servers handling the page requests.

High Availability

IT systems need to be highly available so that they are available when needed.

Scalability

IT systems need to scale as they grow so that increased load can be handled without interruptions to service and redesign of any hardware or software.

Performance

Visitors to a website, for example, do not want to wait more than a few seconds for the pages to load. So it is very important that the performacne of the system is good. Otherwise much potential business could be lost.

Lets assume that we are running a website that gets lots of traffic wevery day. In order to handle all of this traffic the system designer decides that we need 5 servers and we will use a load balancer to distribute web page requests across the 5 servers. Each server has been sized so that it can handle 30% of the requests that arrive.

This means that we only need 3 and a bit servers tio handle all the requests. Why has the designer done this? Is it a waste of money? Well in this case the web company actually makes a ton of money from the traffic to just one server so the fact that 3 servers are busy handling requests is good news for the chief financial officer.

So the reason that the designer has some extra servers is to introduce fault tolerence into the system. This mean that if on server fails there will be no impact on the system and all web page requests will be handled without any loss of money to the business.

Introducing fault tolerance by having extra servers is expensive so the company needs to be sure that it is worth the money. This is typically the responsibility of the capacity planner. He/she can users their knowledge and usually special software to work out how many servers are required to handle a partuicular forecasted load on the system.

This type of load balancing uses part of the DNS functionality that allows more than one IP address to be associated with a single hostname (or server name). n order to load balance with DNS a round robin approach is used where the requests are routed to each backend server in turn. Strictly speaking it is not server load balancing but the DNS system itself runs on servers

The main benefit of this approach is that it is easy to set up and configure. However it is a simple process so there are a number of disadvantages of this approach which include:

• only round robin can be used
• DNS cannot detect if a server breaks down and is not operational
• sticky (same server) connections are not possible
• lod distribution can be unpredictable

The load balancer intercepts the incoming traffic and shares it across the available servers. By doing this the load balancer enables multiple servers to respond as if they were a single device or server. This in turn means that more simultaneous requests can be handled. The number of servers that sit behind the load balancer can be anything from 2 (minimum) to hundreds. The load balancing process is completely transparent to any users (using a website for example).

There are additional benefits to server load balancing. First becasue the load is distributed across multiple servers the service being provided can be highly available. That is if one of the servers break down the load can still be handled by the other servers. A second benefit is scalability. This means that if the load increases substantially more servers can be added behind the load balancer in order to cope with the increased load.

Also note that server load balancing can be achieved using either software running on a server or via dedicated hardware servers.