Big data has not only created new opportunities for industrial and academic research, but also introduced many challenging problems. Collaborative decision making is one of several mechanisms for addressing these challenges. Scalable resolution display environments (SRDEs) have become increasingly popular in the past two decades as a utility for supporting collaborative decision making. Despite this popularity, the concept of a data observatory built using several SRDEs is relatively new. The Data Science Institute of Imperial College London is one of the first institutions to facilitate such an environment. SRDEs require specialised middleware to operate them and only a few dozen variants have been developed so far. It does not come as a surprise that none of these middleware were suitable for the requirements of operating the four distinctive SRDEs found at the institute.

This thesis proposes and validates a novel strategy for scalable rendering of data visualisations using web frameworks. It explores the domain of scalable rendering, introduces various types of SRDEs and thereafter evaluates a number of existing middleware presenting their capabilities and limitations, establishing the key requirements of a middleware for scalable rendering of data visualisations. It proposes that such a middleware needs a combination of a non-invasive application programming model, a modular multi-tier architecture to enable multiple parallel rendering strategies and data distribution through synchronised execution, and a web-based ecosystem for developing composite data visualisations. This hypothesis is validated using a quantitative approach by means of extensive performance and scalability tests and studies on user expectations and usage patterns, and thereafter using a qualitative approach taking three real world projects as case studies. The study is based on two open source middleware frameworks, Tuoris and OVE, which were developed to operate the SRDEs at the institute.