Intro | How links work | Client requests | Server response | Caching | Server-side programmes | Negotiation | Cookies | Logs |Authentication | TCP and IP | FAQ

Caching: agents between the server and client

The browser cache

You may have noticed that when you go back to a page you've looked at not too long before the page loads much quicker. That's because the browser stored a local copy of it when it was first downloaded. These local copies are kept in what's called a cache. Usually one sets a maximum size for the cache and a maximum caching time for documents.

This means that when a new page is visited it is stored in the cache, and if the cache is full (near the maximum size limit) some document that the browser considers unlikely to be visited again soon is deleted to make room. Also, if you go to a page that is stored in the cache the browser may find that you've set 7 days as a the maximum storage time and 8 days have now passed since the last visit, so the page needs to be reloaded.

Exactly how caches work differ between browsers, but this is the basic idea, and it's a good one because it saves both time for the user and network traffic. There are also some HTTP details involved, but they will be covered later.

Proxy caches

Browser caches are a nice feature, but when many users browse from the same site one usually ends up storing the same document in many different caches and refreshing it over and over for different uses. Clearly, this isn't optimal.

The solution is to let the users share a cache, and this is exactly what proxy caches are all about. Browsers still have their local caches, but HTTP requests for documents not in the browser cache are not sent to the server any more, instead they are sent to the proxy cache. If the proxy has the document in its cache it will just return the document (like the browser cache would), and if it doesn't it will submit the request on behalf of the browser, store the result and relay it to the browser.

So the proxy is really a common cache for a number of users and can reduce network traffic rather dramatically. It can also skew log-based statistics badly. :)

A more advanced solution than a single proxy cache is a hierarchy of proxy caches. Imagine a large ISP may have one proxy cache for each part of the country and set up each of the regional proxies to use a national proxy cache instead of going directly to the source web servers. This solution can reduce network traffic even further. More detail on this is linked to in the references.