About GRM agent caches and the groups to which they belong

Function of GRM agent caches

Every GRM deployment includes a set of NetCache appliances that are configured as GRM agent caches. GRM agent caches cache content and resolve content requests on behalf of clients.

When you configure a GRM server, you identify each GRM agent cache and the group with which it is associated. A GRM agent cache is categorized as follows:

An edge cache in an edge group
A center cache in a center group

About GRM edge caches and edge groups

Edge caches are located in groups that are referred to as edge groups. Edge caches in a particular edge group serve content only to the clients associated with that edge group. For example, in a telco CDN, each POP is typically an edge group and each GRM agent cache deployed at the POP is an edge cache. An edge cache in POP A can serve content only to clients in POP A.

Characteristics of edge caches are as follows:

They perform health monitoring and inform the GRM server that they are active. (The GRM server considers health information when it determines which caches are eligible to return content to a client.)
They do not perform proximity measurements. When you configure your GRM server, you specify the CIDR blocks associated with each edge group. The CIDR block subnets associated with an edge group define which clients can be served by caches within the edge group. See Section C, How a GRM server determines IP addresses to return.
They fail over to center caches if no edge caches in the edge group are available. For details, see Failover for GRM servers.

About GRM center caches and center groups

Center caches are located in center groups. A center group is equivalent to a telco or enterprise CDN data center.

Characteristics of center caches are as follows:

They provide content for clients that are not associated with an edge group. For example, if a CDN provides services for POPs and clients not in POPs, the center caches would serve content for clients not in the POPs.
They perform health monitoring, informing the GRM server that they are active.
They measure proximity between themselves and the requesting client (for a GRM L7 server) or the local DNS nameserver (for a GRM DNS server) and return that information to the GRM server.