Tag Caching

In the garbage-collected world, we want to keep the GC overhead as little as possible. One of the things we can do is limiting the number of allocations in our application. How to achieve that? There’s sync.Pool which caches allocated but unused items for later reuse.

The Pool can become very useful when you have multiple parallel operations that can share the same piece of memory between them. The real power of it is visible when you have frequent allocations and deallocations of the same data structure. The Pool scales up when the load increases and can be cleared in every 2 GC circles. I’ll explain the algorithm behind it a bit later. Right now, I focus on how to use it.

The varnish is a well-known HTTP accelerator. As the continuation of the GoInPractice series, today I’ll show how you can build a simple (and naive) varnish implementation in Go. Some of the code is reused from Writing a reverse proxy so if you don’t understand something, I recommend taking a look at the blog post.

We’ll split our project into a few parts. The first one will be the caching mechanism. Its responsibility will be storing the data to cache and invalidate it after reaching the deadline. We’ll use mutexes for synchronization between goroutines.