Modelling Data Centre and IT Infrastructure for Optimised Service Availability, Downtime and Cost

  • Tom Townsend

    Student thesis: Doctoral Thesis

    Abstract

    Technology is becoming increasingly important as digital channels between organisations and consumers are increasingly relied upon. Availability is, therefore, a vital element to consider when deploying and maintaining IT systems. Although its importance is not contested, availability is often considered separately for the data centre and IT infrastructure, which is a problem. This research explores the relationship between IT infrastructure and data centres in the context of service availability.
    The aim of this research project is to explore this relationship and produce a tool for industry decision makers to help weigh their IT infrastructure placement decisions. To this end this thesis takes a design science approach and presents a series of models and recommendations for the deployment of highly available IT systems. Taking a mixed method approach, models are created with their corresponding equations allowing for a structured series of scenarios to be explored. These models were validated and extended through a series of semi-structured interviews with industry experts and decision makers; followed by a thematic analysis. Other elements, such as the cost of downtime, are also taken into account to add an additional dimension to the results.
    The mix of qualitative and quantitative results provides a rich context for the results and allowed for a series of recommendations to industry, which are supported by 2,412 scenarios and interviews results. Of particular interest is the finding that a 68-times improvement in availability can be made to a systems availability, by increasing the number of IT nodes from two to four and evenly distributing those between two data centres. Another key finding relates to high performing scenarios, which used low tier data centres. This has the potential to alter how organisation approach IT deployments and their data centre investments. The cost of downtime was also found to be a critical consideration. The most cost-effective scenarios were significantly different once this was considered.
    As such, this project has generated a tool which provides organisations with a mechanism to make more informed decisions based on what is more important to them; cost or service availability. The models presented provide a simple tool for decision makers to compare different approaches and find optimal deployment configurations, within the constraints of their environment by inputting their own values into the equations provided. The tool is novel in its design as it enables the organisations to work with the infrastructure they already have in place, as opposed to enforcing a rigid structure which may require completely new investments. Therefore, organisations are provided with information required to inform where the most effective investments should be made to achieve the availability goals of the organisation.
    Date of Award2020
    Original languageEnglish
    SupervisorMasoud Mohammadian (Supervisor), Wanli Ma (Supervisor) & Blooma John (Supervisor)

    Cite this

    '