Memory compression has long been a topic confined to academic research and development sandboxes. However, with CPU and memory speed improvements outpacing improvements in I/O speed and latency, memory compression is now being deemed a viable way to increase in-memory data density and delaying or avoiding costly I/O. Zswap, a feature for compressed swap caching, merged into the kernel in v3.11, is a first step toward deploying this functionality in the Linux kernel.
This presentation targets users who own hardware with maxed-out RAM capacities or pay per-GB for RAM usage (IaaS customers) and would like to determine if memory compression can reduce their costs and extend the useful lifetime of their computing assets. Basic knowledge of memory management concepts like memory pages and swapping is a plus, but deep knowledge of the kernel memory manager is not required.