Polly : quantifying the benefits of emerging web technologies over traditional ones in response to upcoming concurrency and real-time challenges

Abstract

With the internet's ever-increasing infiltration in our daily tasks, instant and responsive online experiences are no longer an option but a must. These, paired with scalability requirements and the need to efficiently handle concurrency, sum up the current most dominant web challenges. This study aims to scientifically assess and quantify any discrepancies in performance between an emerging stack of technologies (i.e. event-based web server, HTML5 WebSockets and a NoSQL DB) and a traditional stack (i.e. thread-based HTTP server, long-polling and RDBMS) when addressing the challenges related to trying to support thousands of users interacting with a system while still meeting user expectations (responsiveness and real-time feedback). This is answered through a series of lab-based experiments whereby the strengths and weaknesses of the chosen architectures are evaluated with respect to a specific context (Polly). Polly is a web-based polling application that carries specific scalability requirements with intrinsic user expectations of real-time voting results and it will serve as the environment in which the traditional and emerging architectures are evaluated, against set goals. Results obtained provide many interesting insights on the behaviour of the architectures being evaluated and how these perform under different conditions.