Web-based Real-time Information Dashboard - An event-driven approach to visualize telemetry data

Abstract

Traditional information dashboards where data are updated at fixed intervals or on user interaction do not fulfill all needs for software development practitioners. There is a need for access to information in real-time in order to support decisions in an ever-changing reality. The research presented in this thesis was conducted during 6 months, using a design science research methodology, with the industrial partner Surikat. Part of the study is interviews with three employees at Surikat as well as quantitative performance measurements on a proof-of-concept system developed during the thesis. The proof-of-concept was realized with an event-driven architecture using WebSocket for communication between server and client. Performance measurements revealed that the system gives an average round-trip time, RTT, of 5066ms across all tests. The tests ranged from 1 client and 1 message per second to 50 clients and 80 messages per second, with and without aggregations on the server. The test also showed there are significant difference between browsers. The average RTT was 1640ms for Chrome and 17285ms for Internet Explorer. The proof-of-concept developed during this thesis shows that it is possible to create real-time information dashboards using open source frameworks and emerging web technologies, such as WebSocket. The performance tests show that the system copes well compared to the requirements developed with Surikat but that there are signifi- cant differences between older and newer browsers. In addition, interviews revealed that real-time updates give project managers support to make faster decisions.