A common method of name resolution is to use a method called round-robin. This method maps a single host name to multiple different physical server machines, giving out different IP addresses to different clients. Load balancing is treated in more detail later in this blog (see name resolution methods). With round-robin DNS, the user is unaware of the existence of multiple servers.

The pool of servers appears to be a single logical server as it has only a single name used to access it. Redirections

Another mechanism available for Web servers is to return a redirection to a parallel server to perform load balancing.


For example,

  • upon accessing the URL http: //WWW.mywebsite.com/
  • the main server WWW. mywebsite.com will send an HTTP redirection to URL http: //WWW2. mywebsite.com
  • Another user may be redirected to a different server: http://WWW4. mywebsite.com/

This way, the load can be redirected by the main server WWW to several separate machines Wwwl, WWW2, …, Wwwn. The main server might be set up so that the only thing it does is perform redirections to other servers. There is often a misconception regarding this scheme where it is thought that every request would still have to go through the main server to get redirected to another server.

On the contrary, for any given client, there is only a single initial redirection. After that, all requests go automatically to the new target server, since the links within the HTML text are usually relative to the server where the HTML file actually resides.  It can cause some difficulties in certain situations where there are cached cookies for example, perhaps if you access one of the many BBC servers to watch Match of the Day online like this site.

With this method, the user is aware of the fact that there are several servers, since the URL location field in the client software will display a different server name than originally accessed. This is usually not an issue, though. The entry point to the site is still centralized, through the main server, and that’s the only address they ever have to remember.

However, bear in mind that users may place a bookmark in the client software pointing to one of the several servers sharing the load-—not the main server. This means that once a server name is introduced, say WWW4, there may forever be references to that machine on users’ bookmark files.   Remember that these destinations may be slightly different if the web page is accessed through a bookmark so don’t expect the exact same result.

Although using this round robin method for name resolution is extremely common, don’t assume it’s always deployed.  There are many other methods using variations of this method including different types of redirection or mirroring.

John Hughes


No Comments Networking, Protocols

Leave a Reply

Your email address will not be published. Required fields are marked *