At CES, many researchers distribute their computational intensive tasks to different servers to speed up the execution. Unfortunately, there is no scheduling of resources which often lead to sub-optimal usage, e.g. bad occupation of servers or long run times for urgent tasks.
Your group will design and implement a /task scheduling system/ to distribute tasks among a set of servers. With your application, the user will be able to easily queue a task from the command line. Further parameters provide more options, e.g. setting a priority or defining constraints for the task. After the task has been issued by the user, your application will dispatch the work to a suitable machine and run it when the time is right. Last but not least, a simple webpage shows the occupation of servers and the queue length.
This project might catch your interest if you
- want to develop a Linux application
- like the challenge of taking network communication into account, or
- like to focus on interesting algorithms rather than graphical interface design.
Sprache: Deutsch oder Englisch