Scenario: NetCache deployed at a global carrier and an ISP

About this scenario

This section contains examples of a total Web caching solution for a global carrier and an ISP.

Global carrier's and ISP's goals

The carrier is interested in caching for the following reasons:

The carrier and the ISP want to reduce the use of bandwidth whenever possible.
Fast response of Web pages to the customer provides a competitive advantage.

Global carrier's and ISP's requirements

The global carrier's and ISP's requirements for caching are as follows:

Caching must be invisible to the users; it is not practical to expect the users to configure their Web browsers to send traffic to the Web cache at the ISP.
The caching system must be reliable. Service outages not only inconvenience customers, but the company reputation suffers because the outages are given a good deal of negative attention from the media.

Example 1: all users handled by POPs

The following illustration shows a deployment in which all users are handled by points of presence (POPs).

Details of the preceding illustration are as follows:

By adding NetCache appliances at each level (POP, network hub, carrier), copies of frequently used Web objects are stored locally, thereby reducing network latency and bandwidth consumption. If the object is not in the cache at one level, a NetCache appliance can try to obtain the object from a cache at the next level up.
To fulfill the requirement that caching must be invisible to the user, transparent proxying is deployed at each POP.
A pair of L4 switches, at the same level, were added to the network at the POP to provide redundancy in case one switch fails. This helps to fulfill the requirement for reliability. (A WCCP router could be used instead of L4 switches.)
It is not necessary to deploy transparent proxying at the network hub because all users connect to the POP, where transparent proxying is implemented.
Because the network hub is to handle cache misses at the POP, the NetCache appliance at each POP is configured to view the NetCache appliances at the network hub as a cluster in a logical hierarchy. This configuration enables the NetCache appliances at the POPs to send the requests they cannot resolve to the NetCache appliances at the network hub and distribute the requests among them. See Section A, Request resolution hierarchies for more information.
There is no advantage to deploying transparent proxying at the carrier level because transparent proxying is deployed at the POP level. Instead, the NetCache appliances at the network hub are configured to view the NetCache appliances at the carrier level as a logical cluster. The result is that the network hub's NetCache appliances can distribute requests they cannot resolve across the carrier's appliances.

See Section A, Request resolution hierarchies and the Administration Guide for more information about clusters in hierarchies.

Example 2: some clients connect directly to the ISP, network hub, or data center

In this example, some clients connect to the POPs. Other clients connect to the ISP, network hub, or data center through a T1 line.

The difference between this deployment and the deployment in Example 1 is that transparent proxying is deployed at the ISP, network hub, or data center level to provide transparent services to the clients connecting over the T1 line. The L4 switches are deployed in a pair also, for redundancy, but this strategy is not required. A WCCP router could be used instead of L4 switches.