Overview

Caching hierarchies extend an appliance's cache performance by enabling it to get uncached objects from other appliances on the network rather than from the origin Web servers on the Internet. The mechanism for achieving communication between proxy servers is the caching hierarchy.

Caching hierarchies make it possible for a high percentage of browser requests to be filled from within the network.

The Excelerator appliance supports two types of caching hierarchies: CERN and ICP.


CERN Hierarchies


Key Functionality

CERN hierarchies consist of CERN clients and CERN parents. The client is basically an appliance that is configured to use one or more proxy servers as forward proxy servers (CERN parents).

CERN hierarchies are very similar to the relationship between a client workstation browser and a forward proxy server. The difference is that the CERN client is an intermediary and the browser is a user agent.

CERN parents can also designate other proxy servers as their CERN parents.


How CERN Hierarchies Work

Figure 40


Benefits of CERN Hierarchies

For tips and guidelines on setting up CERN hierarchies, see CERN Hierarchy Setup.


ICP Hierarchies


Key Functionality

ICP hierarchies consist of ICP clients, parents, and peers. The ICP client is an appliance that is configured to communicate with one or more proxy servers that have been configured as ICP servers. These ICP servers are designated as either peers or parents of the ICP client.

Servers in an ICP hierarchy use:

ICP hierarchies are more complex than CERN hierarchies. The ICP protocol offers intelligence for deciding which routes to pursue first when seeking data.

ICP parents can also specify other proxy servers as their ICP parents or peers.


How ICP Hierarchies Work

NOTE:  Steps shown in reverse color are conditional.

Figure 41


Benefits of ICP Hierarchies

For tips and guidelines on setting up ICP hierarchies, see ICP Hierarchy Setup.