Implement new scheduling algorithms in Haizea

The Haizea Lease Manager can be used as a drop-in replacement for OpenNebula's scheduling daemon. Haizea's scheduling code currently supports scheduling VMs for best-effort leases, advance reservation leases, and immediate leases (see the Haizea website for more details on what these leases are). However, there are other types of interesting leases that could be supported in Haizea: deadline-sensitive leases (which can be scheduled on a best-effort basis, but have to be completed before a specified deadline), lease futures (where a user wants guaranteed resources at some point in a known time interval, but doesn't know exactly when), etc. The scheduling problem becomes more interesting when even more policy-driven constraints are added (a policy engine will be added to the next version of Haizea and a preliminary version will be available before GSoC starts). The project goals are open-ended, and it is up to the student to propose specific scheduling algorithms or new types of leases he/she wants to implement and experiment with. This project can be an excellent learning experience for students (specially at a graduate or advanced undergraduate level) who are interested in resource management, VM scheduling, etc.

To find out more about OpenNebula visit OpenNebula Management Console project and project site

Required Skills
Basic understanding of scheduling algorithms is desired. For developement intermediate level skills requird on Python programming besides working knowledge of OpenNebula, Xen virtualization .