Can’t believe anyone is STILL advocating this.

Edge servers are automatically behaving like Ram disk – only much, MUCH better; it uses the memory dynamically and in a way that’s much manageable by the OS. Unlike Ram disk who’s taking away the memory from the OS.

And edge servers WILL fallback to disk cache gracefully – therefore still giving excellent performance boost (because the request still doesn’t hit the system). Unlike a Ram disk.

Edge server – also known as reverse-proxy server, provides the MOST significant scalability & performance boost, by avoiding most requests to even hit the system.

Most of the requests won’t even hit the webserver. Many will be served straight from memory. Some more from the disk cache (read: static files, further cached by the operating system)

In a case, a system was serving 2 requests per second at maximum, with 100% processor load. After deploying an edge server, it handles 500 requests per second, with 30% processor load.

The performance gain alone is simply mind-boggling. And it is scalable to the max – you can deploy multiple edge server, spreading the load between them.

This is the top mistake, most definitely. Even so-called “experts” misses this : You START from the edge. From then you continue to the center of the architecture.