Scaling Parallelism Under CPU - Intensive Loads in Node.js

An increasing number of applications are using Node.js, a framework for asynchronous I/O, event-driven, server-side JavaScript. The backbone of Node.js is the single-threaded event loop. Therefore, computationally intensive tasks are bound to the performance of a single core. Modules with different characteristics have been built to provide parallelism and scaling. We evaluate the performance of some representative Node.js multiprocess and multi-thread techniques focusing on their scaling behavior on different environments. We present computation metrics using a compute-intensive task as a constant. Finally, we use statistical analysis to identify similarities and differences in performance with the end goal of providing recommendations on deployment.