Abstract

To oblige with the legal requirements, organizations have to keep data up to a certain amount of time.They are creating a huge amount of data on daily basis therefore it is very difficult for them to manage and store this data due to the legal requirements. This is where Enterprise Content Management (ECM) system comes into picture. ECM is a means of organizing and storing an organization's documents and other content that relates to the organization's processes. With ECM being offered as a service, thanks to cloud computing it makes sense to offer this functionality as a shared service. There are various benefits of offering it as a shared service one of which is that it is a cheaper method to meet the needs of large organizations with different requirements for ECM functionality. ECM systems use resources like memory, central processing unit (CPU) and disk which are shared among different clients (organizations). With every client, a service level agreement is there which describes the performance criteria a provider promises to meet while delivering the ECM service. To improve the performance of the ECM by optimizing the use of resources and match the Service level agreements various techniques are used. In this thesis, heuristics technique is used. Performance baselines and utilization of resources are measured for different workloads of the clients and on the basis of that, resources of the ECM can be dynamically provisioned or assigned to different clients get the optimized resource utilization and better performance. First of all typical workload is designed which is similar to the work being performed by various banks and insurance companies using IBM ECM systems and which consists of interactive and batch type of operations. Performance baselines are being measured for these workloads by monitoring the key performance indicators (KPIs) with variable number of users performing operations on the system at the same time. After getting the results for KPIs and resource utilization, resources are being assigned dynamically according to their utilization in a way that the use of resources is optimized and clients are satisfied with better service at the same time.