Energy and Runtime Performance Optimization of Node.js Web Requests

Abstract

The Node.js framework uses an event-driven model with a single-threaded event loop and provides asynchronous and non-blocking I/O operations. As with other programs, Node.js web applications take advantage of underlying resources, including CPUs, which can incorporate the dynamic voltage and frequency scaling (DVFS) technique. Using CPU DVFS, the applications can increase their runtime performance, at the expense of the system's energy consumption. Thus, software code that utilizes the CPU DVFS technique efficiently should lead to “green” and high-performing applications with respect to the business logic. To this end, we build a CPU frequency scaling/energy aware system to enable CPU frequency control within Node.js applications and measure the energy consumption of specific tasks. We also build a benchmark suite to analyze the energy consumption and runtime performance of different requests based on the CPU frequency impact and collect information and patterns, as we scale the CPU frequency. The analysis aims to provide data and knowledge on the CPU frequency “suitability” and impact in order to create a model for CPU frequency scaling on Node.js web applications and achieve an efficient and sustainable runtime performance.