HostingGreen web hosting packages
TutorialsBeginners guides to SEO & hosting
Beginners guide to Web Hosting
Round robin DNS configuration
Multiple Name Servers: As described in the previous tutorial on, two to seven name servers can be registered against a name. These are DNS servers that contain information on where to find services (such as www or email) for a particular domain name. There are two main reasons for providing more than one DNS server for each domain. The first is to provide protection against a DNS server going down- if a DNS server failed for any reason (eg. power outage, communication failure etc.) the using this DNS server would be unobtainable since the of these domains could not be retrieved. Instead, having two or more (which should be geographically and topologically seperate) means that if a request to one of them fails, the name server can be requested again. The name servers are requested and in normal circumstances, information from any of the DNS servers is used (which normally contain the same information). The chances of two DNS servers going down independently to each other are very slim.
Lets walk through what happens when one DNS server fails for a request to www.-hat.com. The name servers registered against the domain name are “ns1.wight-hat.com 22.214.171.124″ and “ns2.wight-hat.com 126.96.36.199″ when one of the name servers is down.
- The ISP requests the name server for wight-hat.com and receives the name servers. The name servers are “ns1.wight-hat.com 188.8.131.52″ and “ns2.wight-hat.com 184.108.40.206″.
- A request is sent to 220.127.116.11 for the location of wight-hat.com but, this DNS server is not online and the request fails.
- A request is sent to 18.104.22.168 for the location of wight-hat.com and the DNS record information is returned.
Round Robin configuration: Each DNS record can have more than one record for a given domain. For instance, when looking for a web page the DNS supplies an “A” record but there can be more than one A record entry. These IP addresses are delivered in rotation so if there were two A records, alternate IP addresses would be served on each request. Suppose a very high traffic website was starting to show a speed lag because the server that the site was on could not handle the amount of requests/ traffic/ processing. An additional server could be employed so that it would share the load. So how do you share the load when you have one domain name that everyone is accessing the website through? You configure one DNS records to point to both the first web server and second server. In this way, each request will receive alternating IP addresses and these will give alternating IP addresses for the website servers thus sharing the load 50/50. With up to seven name server entires it is simple to configure up to seven web servers to share the load of a very high traffic website. This is known as load balancing.
If one of the servers had twice the capacity of the other, you may wish to send it twice as much traffic. In this case you would duplicate the A record for the high capacity server so that you had 3 A records- 2 for the high capacity server and 1 for the other server. Your server load would now be split 66/33.
Whilst this is an effective way to load balance servers, it provides no protection against a server becoming unavailable. Using DNS records we have come up with a method of providing protection against this eventuality as well as load balancing which we discuss in the next section.
Next section: Load balanced and redundant server network
Index of all our tutorials: Wight Hat tutorials