Agnostic AJAX: Asynchronous JavaScript and Data

We present an experimental study comparing the relative performances of a traditional real-life production HTML application and a set of comparable AJAX applications that implement the same user interface. These applications use several different data methodologies to perform the AJAX updates: partial HTML, XML, JSON, and CSV (comma separated values). A significant number of queries on a set of tasks were used to collect data on the performance of each update methodology. These tasks covered a wide range of response sizes. We discuss the performance of each methology, the advantages and disadvantages and the strengths and weaknesses for both developers and clients. We also comment on the applicability of these technologies for special cases, such as providing data on the go for mobile clients.