JANET, the UK academic network, has a dedicated undersea cable connection to the USA known as the Fat Pipe. All UK higher education institutions access American Internet sites via the Fat Pipe, and at certain times this makes connections slow down. To make matters worse, there is no direct JANET connection to the Pacific area (Australia, New Zealand, Japan etc): any connection requests to that part of the world first travel to the US via the Fat Pipe, then cross the USA and press on to their destination.
One way round this problem is to keep local copies of frequently accessed documents and files. One way to do this with archives is to set up mirror sites, as described in the previous question, but not all sites are mirrored. An alternative is to set up a proxy or cacheing server and send connection requests via the server rather than directly.
Essentially a cache is used for holding temporary copies of something; in the ftp context, the cache holds files retrieved from ftp servers. The sole purpose of a cache is to save time - and more importantly, bandwidth - if the file is requested a second time. Let's suppose that yesterday morning I downloaded some software from an archive in the USA, and later in the day you tried to retrieve the same package from the same location. It took me a while to retrieve the file, and it would probably take you even longer in the afternoon, as the transatlantic link tends to be busier then. If we had both configured our browsers to use the cache, a copy of the file would have been placed there at the same time as I retrieved it to my computer. When you tried to retrieve the file later, your browser would first check whether there was a copy of the file held on the cache, and discovering that there was, would retrieve it from there rather than from the original server. As you got the file from a machine on the King's network you would normally save a significant amount of time.
There are several levels of caching. Browsers usually have caches; Netscape, for example, sets aside disk space by default in order to save copies of the pages you download. A browser cache is only accessible to the user(s) of that browser, but servers can also have caches: they permit a large store of material to be held at the server site, for all local users to access. At a higher level, we can have national caches; the national cache at HENSA has a vast capacity and holds copies of pages downloaded by any user in the UK who chooses to make use of the cache.
If a user configures their copy of Netscape (or other browser) to point to the local cache here at King's, and requests a file, the browser will first check its own cache, then the local server cache. If the page isn't found, the cache server will contact the parent and neighbours. If a parent or neighbour cache does not reply within 2 seconds, then the request will be forwarded to the default source.
