Characteristics of multithreading models for high-performance IO driven network applications

TL;DR

This paper evaluates different multithreading models for high-performance network servers on modern multi-core systems, providing a roadmap for optimal task distribution to enhance efficiency.

Contribution

It creates, implements, and assesses various task distribution models for multithreaded network applications on current hardware and operating systems.

Findings

Different models exhibit varying efficiency depending on workload and hardware.

The paper offers a comparative analysis of multithreading strategies.

Provides guidelines for selecting multithreading models in server applications.

Abstract

In a technological landscape that is quickly moving toward dense multi-CPU and multi-core computer systems, where using multithreading is an increasingly popular application design decision, it is important to choose a proper model for distributing tasks across multiple threads that will result in the best efficiency for the application and the system as a whole. The work described in this paper creates, implements and evaluates various models of distributing tasks to CPU threads and investigates their characteristics for use in modern high-performance network servers. The results presented here comprise a roadmap of models for building multithreaded server applications for modern server hardware and Unix-like operating systems.