Domains, DNS & Certificates
6 curated resources, easiest first.
A plain-English explanation: a load balancer spreads requests across several servers, while a reverse proxy is a broader tool that can load-balance and also cache, terminate SSL, compress and filter traffic. Often one product does both jobs.
With one public IP and a domain, run a single nginx reverse proxy that sends device1.example.com, device2.example.com and so on to different internal servers by name. Only that one proxy needs ports forwarded. Anything exposed this way should be secured (HTTPS, authentication) or kept behind a VPN instead.
Explains the benefits of a reverse proxy in front of an application server, focusing on slow clients: nginx buffers slow connections so the limited app workers stay free. It also covers SSL handling and absorbing traffic spikes.
Proxies rewrite headers such as redirects, but not the links inside the page, so an app that uses absolute paths (/css/style.css) breaks under /subfolder/. The answer lists the real options: make the app use relative paths, deploy it under the same sub-folder internally, or rewrite page content with a filter module as a last resort.
When traffic passes through more than one proxy (for example a CDN, then a load balancer), nginx logs the wrong client address. The answer explains the realip module's real_ip_recursive option, which walks the X-Forwarded-For chain past your trusted proxies to the real client.
To publish a backend at /foo/ without redirecting the browser, use a transparent rewrite inside the location block (rewrite ... break) and then proxy_pass. A 301/302 redirect would send visitors to an internal address they can't reach. The answer also suggests testing with curl -i.