Release notes

Entropy 1.2.0

The version 1.2.0 of Entropy is a generalization of the dynamic management of VM in clusters. Leveraging the version 1.0.0, this new research prototype aims at providing a way to define advanced scheduling policies for virtualized jobs (that is jobs encapsulating into VMs) on a cluster. The major advantage in comparison with the previous release concerns the used of run, stop, suspend and resume VM capabilities in addition to the migrate one. Administrators can now implement their own scheduler objectives to dynamically manage under-used and overloaded clusters. The code is under heavy development and cannot be considered as stable.

  • Isolation of the different parts of the code
  • Generalization of the VMRP module into the Context Switch resolution problem.
  • Implementation of a default decision module which aims at performing a maximum of virtualized jobs.

Entropy 1.0.0

The version 1.0.0 of Entropy is a complete rewrite from scratch of a previous research prototype. This has been made to clean the code and improve its usability. This release is the first one so the notes only summarizes some major points:

  • Isolation of the different parts of the code
  • Implementation of a default control loop that perform dynamic consolidation
  • Implementation of a adapter for the Ganglia monitoring system
  • Implementation of a migration driver for Xen 3.2
  • A number of SLOC divided by 3, replaced by documentations and tests
  • A small set of documents focused on the deployment and the integration of Entropy into an existing cluster environment
  • Entropy is published onto the LGPL license.