Edge DNS GSLB

A long time ago, when the internet was far smaller, IP addresses were used to reach services. The introduction of qualified names for the services has simplified usage, particularly when it comes to browsers. DNS has been created to manage the hierarchy of names and fully qualified domain names which appear on the web to name all the services. We still use today the base principles defined at the early stage of the Internet.

Destination

DNS is used quite often, in the background, as an easy solution to get the technical information required for reaching an application or a service. DNS gives the destination information, in this case an IP address, potentially a port number and for some more specific usages it can provide rich information like signature or authority proof. Once the IP address is known, a TCP or UDP session can be established in order to transport the data.

Recursion and Caching

For efficiency purposes, the DNS service is hierarchical, distributed and resilient. It uses a smart principle of recursivity in order to ask the server in charge (authoritative server) for the next information required to reach the ultimate goal: getting the IP address of an FQDN. This recursion principle is used on all IP networks – private and as an extension on the whole Internet. The recursion is associated with a caching system which prevents repeated asking of the same information during a period of time considered as small enough for the information to remain valid.

Application Routing

In order to access an application from a client or from another application component, the DNS service is used to convert FQDN into an IP address and optionally gather other technical information. But DNS is not the only component contributing to application access. Of course, the IP network has a massive role to play: it routes IP packets from source to destination and is quite complex. On the path, we can also find security components and more importantly, for applications, some load-balancing solutions. The load-balancers are generally used to distribute the application sessions across multiple servers that can therefore handle load and some failure situations. As part of this overall routing process the DNS plays the major role of «defining the destination», this is why its role is critical and the impacts of the DNS service failing are huge.