Proxies: What's the difference?
You've probably heard of proxies. Perhaps you used one when you were at school to access a website you weren't supposed to. But did you know that there are multiple different types of proxies that are used for different things? For example, a reverse proxy perform load-balancing and caching for your web application? And that a transparent proxy can be used to filter the traffic of your internet connection without you knowing (well, almost)? In this post, I'll be explaining the difference between the different types of proxy I'm aware of, why you'd want one, and how to detect their presence.
Reverse proxies also come in really handy if you want to run multiple, perhaps unrelated, servers on a single machine with a single IP address, as they can reverse proxy requests to the right place based on the requested subdomain. For example, on my server I not only serve my website (which in and of itself reverse-proxies PHP requests), but also serves my git server - which is a separate process listening on a different port behind my firewall.
Caching is another key feature of reverse proxies that comes in dead useful if you're running a medium-high traffic website. Instead of forwarding every single request to your backend for processing, if you've got a blog, for instance, you could cache the responses to requests for the posts themselves and serve them directly from the reverse proxy, leaving the slower backend free to process comments that people make, for example. Both nginx and Varnish have support for this. This with method, it's possible to serve 1000s of requests a minute from a very modestly sized virtual machine (say, 512MB RAM, 1 CPU) if configured correctly. Take that, Apache!
Finally, when 1 server isn't enough any more, your can get reverse proxies like nginx to act as a load balancer. In this scenario, there are multiple backend servers (probably running on different machines, with a fast internal LAN connecting them all), and a single front-facing load balancer sitting in front of them all distributing requests to the backend servers. nginx in particular can get very fancy with the logic here, should you need that kind of control. It can even monitor the health of the backend application servers, and avoid sending any requests to unresponsive servers - giving them time to recover from a crash.
Forward proxies are distinctly different to reverse proxies, in that they make requests to the destination client wants to connect to on their behalf. Such a proxy can be instituted for many reasons. Sometimes, it's for security reasons - for example to ensure that all those connecting to a backend local network are authenticated (authentication with a forward proxy is done via a set of special
Proxy- HTTP headers). Other times, it's to preserve data on limited and/or expensive internet connections.
More often though, it's to censor and surveil the internet connection of the users on a network - and also to bypass such censoring. It is in this manner that HTTP(S) has become so pervasive - in that companies, institutions, (and, in rare cases), Internet Service Providers install forward proxies to censor the connections of their users - as such proxies usually only understand HTTP and HTTPS (clients request that a forward proxy retrieve something for them via a
GET https://bobsrockets.net/ HTTP/1.1 request for example). If you're curious though, some forward proxies these days support the
CONNECT HTTP method, allowing one to set up a TLS connection with another server (whether that be an
SMTPS, or other protocol server). In addition, the
SOCKS protocol now allows for arbitrary TCP connection to be proxied through as well.
Forward proxies nearly always require some client-side configuration. If you've wondered what the proxy settings are in your operating system and web browser's settings - this is what they're for.
Such can usually by identified by the
Via and other headers that they attach to outgoing requests, as per RFC 2616. Online tools exist that exploit this - allowing you to detect whether such a proxy exists.
Transparent proxies are similar to forward proxies, but do not require any client-side configuration. Instead, they utilise clever networking tricks to intercept network traffic being sent to and from the clients on a network. In this manner, they can cache responses, filter content, and protect the users from attacks without the client necessarily being aware of their existence.
It is important to note here though that utilising a proxy is by no means a substitute for maintaining proper defences on your own computer, such as installing and configuring a firewall, ensuring your system has all the latest updates, and, if you're running windows, ensuring you have an antivirus program installing and running (Windows 10 comes with one automatically these days).
Even though they don't usually attach the
Via header (as they are supposed to), such proxies can usually be detected by cleverly designed tests that exploit their tendency to cache requests, thankfully.
So there you have it. We've taken a look at Forward proxies, and the benefits (and drawbacks) they can provide to users. We've also investigated Transparent proxies, and how to detect them. Finally, we've looked at Reverse proxies and the advantages they can provide to enable you to scale and structure your next great web (and other protocol! Nginx supports all sorts of other protocols besides HTTP(S)) application.