Thursday, August 4, 2011

Deploying CACHEBOX


Forward caches
Also known as forward proxy caches, they are usually deployed by Internet service providers (ISPs), schools and corporations to save bandwidth.  In order to use the cache in this way you need to configure individual browsers with the network settings of the cache.
In the following example a CACHEBOX is deployed on the same network as the users.  It can in fact be deployed anywhere on the LAN, WAN or Internet, however it generally makes sense to have it as close to users as possible.  LAN speeds are usually much faster than WANs so files will be sent to users quicker giving them a much better user experience.
Basic forward
       
A transparent forward cache, also known as an interception proxy caches, is a type of forward proxy that doesn’t require clients to be configured to use it. There are multiple ways of achieving this, including having a router/firewall send traffic to a forward cache or placing the cache directly in-line.

Gateway mode
In the example below the clients have their default gateway set as the CACHEBOX, which has its default gateway set as a NAT firewall. The CACHEBOX redirects HTTP traffic via its cache and passes the rest directly on to the NAT firewall.
However, because it introduces a single point of failure into the network this type of deployment is generally discouraged.  If it is used it is only really suitable for small networks.
Inline deployment

WCCP deployment
In the example below a Cisco router ISP is configured to separate out web traffic automatically and forward it transparently via a cluster of CACHEBOXs, using WCCP.

WCCP deployment
Reverse caches
Also known as gateway caches, surrogate caches, or web accelerators, reverse caches are deployed in front of web servers.  They operate on behalf of the origin or content server, and are essentially invisible to clients. A number of gateway caches can work together to form a Content Delivery Network (CDN.

The following diagram shows a typical reverse caching set up, using a cluster of CACHEBOXs and three content servers:
reverse caching
Because most of the queries are served from the CACHEBOX cluster it reduces the load on the content servers.