Caching proxy
A forward proxy that stores copies of responses and serves them again, built for saving bandwidth rather than for anonymity.
A caching proxy exists to answer the same question twice without asking the origin twice. It sits between clients and the internet, keeps copies of the responses that pass through it, and serves a stored copy when the same resource is requested again. The purpose is efficiency, saving bandwidth and cutting latency for content many people fetch, which is a different goal from the anonymity most of this glossary is about.
Squid is the canonical example, and it is worth naming because it explains a pattern you meet constantly on free proxy lists. Squid is caching-proxy software that organisations run internally, and its default port, 3128, turns up over and over on public lists precisely because so many Squid installations get exposed to the internet by accident. A large share of the open proxies described elsewhere in this glossary are misconfigured caching proxies rather than services anyone meant to publish.
The mechanics are governed by the same HTTP cache rules browsers follow. Responses carry directives saying whether they may be stored and for how long, and a well-behaved caching proxy honours them, revalidating stale entries rather than serving them blindly. That is also its limitation for encrypted traffic: it cannot cache inside an HTTPS stream it is only tunnelling, so on the modern web its caching role is far narrower than it was when most traffic was plain HTTP.
For anonymity work it is mostly a thing to understand rather than use. An intercepting caching proxy on a network path is a transparent proxy by another name, forwarding your real address while it caches. A caching proxy you deliberately configure is a forward proxy whose extra behaviour is storage. Either way, caching is about serving content efficiently, and any effect on how you appear to a destination is incidental and usually unwanted.
Frequently asked questions
What is a caching proxy used for?
Saving bandwidth and reducing latency by storing responses and serving repeat requests from the copy instead of fetching from the origin again. Organisations run them so many users pulling the same content cost the network once rather than once each. It is an efficiency tool, not an anonymity tool, and any effect on how you appear online is incidental.
Is Squid a caching proxy?
Yes, and it is the best known one. Squid is caching-proxy software widely deployed inside organisations, and its default port 3128 appears constantly on free proxy lists because so many installations are accidentally exposed to the internet. Many of the open proxies you find are misconfigured Squid caches rather than intentional public services.
Can a caching proxy cache HTTPS content?
Not the contents of a tunnel it is only relaying, since it cannot read inside TLS without breaking the encryption. It can cache plain HTTP responses and can cache HTTPS only in an intercepting setup that terminates and re-signs TLS, which requires installing its certificate on your machine. On the mostly-encrypted modern web, its caching role is much narrower than it once was.
Back to the full glossary.