Maximization of resource utilization through dynamic provisioning and de-provisioning in the cloud

The amount of data companies have to manage daily increases drastically from year to year. Whenever work is done information gets created related to a business process. Companies are legally bound to keep record of all their business activities to be able to provide evidence in case of lawsuits. In case of a lawsuit all data relevant to the case has to be collected, put on hold, be reviewed and produced. Lawsuits that require data to be discovered are not rare. According to survey of Fulbright & Jaworski L.L.P. in 2010 16% of the consulted companies had over 50 lawsuites commenced against them. To reduce the amount of documents that have to be reviewed, documents should not be kept longer than the company is legally bound to. But this management is not an easy task. Companies need to introduce big content management systems that need to cope with great amounts of data. To do so huge expenses into IT infrastructure have to be made and complex systems need to be managed. With the advent of the cloud computing paradigm the costs of these software systems could be decreased drastically. IBM is working on a project to provide Electronic Archiving Management as a service using existing Electronic Content Management systems. This service provides means to reliably store documents that are not used in active business processes anymore. Documents are stored until the customer is not legally bound to keep the documents anymore. They are being full text indexed to be easily discovered for litigation cases. This thesis gives an introduction into Electronic Content Management, Cloud Computing and Multitenancy to provide the basics for this project. After introducing a modeling concept based on Tivoli Service Automation Manager's Service Definitions. The existing software components are examined on their capabilities on multitenancy and segregated into manageable resources. The dependencies of the software components are then being examined and an automatic deployment process is developed. With this deployment process it is possible to reduce the time needed to deploy a new installation from weeks to less than two hours.