Why Cache?

Not all proxies are equal. Many, but not all have the ability to cache web pages and objects. Web Caching is the act of storing copies of Web pages on a "local'' system (in our example, the proxy). If the same pages are requested at a later time by the same or other users through the proxy, and the cached copy is still valid, there is no need to contact the original server again. Cache hits can significantly reduce latencies and network bandwidth usage.

All that sounds great, especially the part about reducing latency and saving bandwidth, but what is the risk (if any) to caching? The obvious risk is data becoming outdated. If a cached version is different than what's on the webserver, the person requesting the data isn't getting the most up-to-date information. In this instance it's important to either obey the web page (if it's marked to not cache information on the page), or have a proxy that uses an intelligent algorithm to verify "freshness" of data. One major proxy vendor claims to have a patented method around verifying the "freshness" of their data to keep the most up-to-date information in cache, while keeping bandwidth usage to a minimum.

A truly flexible proxy should let you set your own policy around how long to cache different types of data. Perhaps you want to override settings found on a specific webpage, like Youtube, so the next time a really popular video gets forwarded around the company you aren't downloading it from the web everytime (per the default cache settings on Youtube), and instead cache Youtube video for a few hours or more.

Secure Computing finally added a cache to their Webwasher appliance last year (2007), but even in the latest presentation I heard them give, they generally don't recommend you turn it on, with the main concern around staleness of data.

Caching can definitely work to your advantage, just make sure the proxy you use has up-to-date caching technology to help you set the policy you need.