Measuring Consistency Metric for Web Applications

Web application performance analysis is critical in determination of content delivery time delay and user experience issues. Typical approaches focus on a nominal performance perspective where individual components of web application resource delivery is analyzed and optimized over a presumed minimum delay baseline such as the propagation delay for given distance traveled over the network. Each component of the system is bound to cross various administrative domains and technical demarcation points resulting in a tendency towards an over-provisioned solution set. On the other hand, the user experience has a subjective component where the consistency experiences may dictate a lesser performance as acceptable for an application. That is, if user experiences are taken as the nominal optimization constraint, the baseline measurement of the consistency experienced by the user may serve the analysis and the consecutive improvement measures both more useful and maybe more importantly, more cost-effective. In this respect, this paper presents a novel baseline measurement metric called a consistency metric to utilize as a baseline measurement parameter in making optimization improvements on the web application performance. We present our method of measurements and determination of the consistency metric using our highly distributed geographically-diverse data collection system, NetForager, using two popular applications CNN and YouTube.