Section A: Deployment with only a GRM DNS server

When GRM DNS servers are used

The following illustration shows a deployment in which only a GRM DNS server is used for redirection services. Deploying one or more GRM DNS servers is appropriate in the following circumstances:

When clients are physically close to their local DNS nameservers (as in this example), you can deploy only GRM DNS servers to provide redirection services (a pure GRM server deployment).
You choose to redirect client requests to a GRM L7 server by using a GRM DNS server. See Requirement for directing client requests to a GRM L7 server for more information.

The preceding illustration shows the resolution of a DNS query from a client in a POP. Both the GRM DNS server and the back-end nameserver for the GRM DNS server are located in the CDN data center. A back-end nameserver can be in a location different from the GRM DNS server, and the GRM DNS server can point to it, however. For simplicity, the example shows only one GRM DNS server and one back-end DNS nameserver. Typically, multiple GRM DNS servers and multiple back-end DNS nameservers would be deployed for redundancy.

A GRM DNS server can assume the identity of the existing authoritative DNS nameserver for the CDN name space; that is, the GRM DNS server can assume the IP address that was previously assigned to the authoritative DNS nameserver. The authoritative DNS nameserver is then reconfigured with a new IP address. The authoritative DNS nameserver becomes the back-end to the GRM DNS server in this case.

If the GRM DNS server returns multiple IP addresses of GRM agent caches, the client browser selects one GRM agent cache from the list returned and sends the content request to that cache. As the previous illustration shows, if a GRM agent cache does not have the requested content (a cache miss), the cache fetches it from the CDN origin server.