User Interface to Control and Visualize the Planning of Resource Allocations using Genetic Algorithms (Bachelor Thesis, Finished)


Thomas Ritter


Cloud Computing, or in general Service-Oriented Architectures (SOAs), have attracted a lot of interest in the recent past because of the promise of virtually unlimited resources: for example, by distributing computationally complex applications onto multiple resources, end users could experience a significant overall speedup for their calculations.

Scientific Workflows are particularly demanding and resource-hungry for an infrastructure, because they involve both long-running operations and large data transfers. An approach that allows to maximize utility for consumers, in terms of their demands, as well as providers, in terms of their provision capabilities, involves the planning of Advance Reservations in order to schedule the availability of resources, thus improving predictability for both sides, while considering the state of the infrastructure and user-specified Quality of Service (QoS) requirements.

However, the actual optimization algorithms employed for the planning are only one side of the equation. It is equally important to enable end users (and developers!) to steer and supervise the planning process itself, by providing them with an easy-to-interpret interface.

The UI enables non-expert users to configure the optimization criteria by specifying their QoS demands in an intuitive manner, without necessarily needing to know about complex details. In addition, it allows users to follow and gauge the optimization process as it is proceeding, providing various useful levels of information. The goal is for the user interface to be simple and clear, with the possibility to obtain details on demand.

Start / End Dates

2010/10/04 - 2011/01/31


Research Topics