Characterizing the delay performance of web-based cloud services in a browser-scripting approach

Cloud services are often provisioned to their customers using user-friendly web browsers with flexible and rich plug-in environments. Delay is one of the fundamental performance metrics of these web-based services. Commonly-used network measurement tools usually only measure network delay and it may be difficult to infer the web-delay performance using only network layer measurement approaches. In this paper, we propose to characterize the application layer delay in a browser-based network measurement platform using scripts. We conducted a delay measurement study using engineered scripts in the proposed browser-based measurement platform. Our investigation included a comparison study of three browser-scripting delay measurement methods, including Java applet, JSP and Flash ActionScript. We developed a browser-based delay measurement testbed over the Internet so that different delay measurement tools could be evaluated in the same real network environment. We also decomposed the components of the end-to-end delay process of the above measurements to reveal the difference and relationship between network-layer delay and application-layer delay. Our measurement results characterize the stochastic properties of the application layer delay over real Internet paths, and how these properties varied from the underlying network layer delay. The browser-scripting measurement approach can be easily cooperated with different cloud service platforms to inspect their application-layer delay performance. The results may provide insights into designing new cloud services with enhanced quality-of-experience perceived by cloud users.