You are here

Quality of Experience for web browsing

General description

A variety of factors may lead users to a poor quality of experience in a web browsing session, experiencing a high page load time.

Examples are given in figure. 

A high page load time can be caused by:

  • An overloaded client;
  • A congestion in the local area network to which the client is connected;
  • An overloaded web server, serving the web page,
  • Any secondary server providing content to the web page,
  • A congested link somewhere in the path between the user and the web server.

This use case demonstrates how the mPlane can be used to monitor and find the root causes of end-user quality of experience degradation in Web browsing. To achieve this goal, we browse web pages performing active and passive measurements and we collect all the network level information needed to run a diagnosis algorithm identifying the root cause of a high page load time. Furthermore, each probe sends to a Hadoop Distributed File System (HDFS) repository the aggregated data carrying also its own ``vision'' of the network, and the Reasoner pinpoints the cause of performance degradation on specific targets (i.e., web pages) by exploiting data from different Firelog mPlane probes.


Components describing the architecture, and point to pieces of software

The core component of this use case is the Firelog probe, instrumented to receive a mPlane specification containing a url and to return a local diagnosis on the current web browsing session, after sending its measurements to the HDFS repository.

The reasoner for this use case exports the capability to diagnose a url. Given a probe identifier and a url, the reasoner exploits the data collected by different vantage points to run an enhanced version of the diagnosis algorithm, having as an input also the results of specific analysis modules run at the repository side.


How to setup and deploy the use-case  


For the detailed instructions on the deployment, setup, and demo of the use-case we refer interested readers to the corresponding demontration guidelines page (link).