Profiling user behaviour for efficient and resilient cloud management
Cathryn Peoples, Gerard P. Parr, B. W. Scotney, Sanat Sarangi, Subrat Kar · 2014
User behaviour profiling can be used within network management schemes to indicate the capabilities required from management proxies in terms of the way(s) and rate at which they should be aware of real-time network state and change in resource demands/requirements. A management proxy in a cloud, for example, needs ability to monitor change in the most popular webpages associated with a website or files associated with an application so that this detail may influence the caching strategy for optimised performance and operation. When resources are provisioned dynamically across a cloud, the strategy to do so will accommodate efficiency and security objectives, and also take into account the ways in which users are demanding services to optimise the opportunities that their requirements are met. Many online companies now operate in this way and analyse customer behaviour to improve services by meeting predictable requirements and utilising unpredictable behaviour. It is therefore to this gap we respond in this work. A network management model is developed around behaviour profiles of user activities associated with the Wireless Sensor Knowledge Archive (Wisekar) website hosted in the Indian Institute of Technology in Delhi, India. Trends in user access and activities with the website are identified and, in response, a management framework is proposed. A Certainty Factor quantifies the confidence with which management is applied such that actions enforced accommodate both predictable and unpredictable user behaviour.