ENTICE: dEcentralized repositories for traNsparent and efficienT vIrtual maChine opErations
In this project, we research and create a novel VM repository and operational environment for federated Cloud infrastructures aiming to:
simplify the creation of lightweight and highly optimised VM images tuned for functional descriptions of applications;
automatically decompose and distribute VM images based on multi-objective optimisation and a knowledge base and reasoning infrastructure to meet application runtime requirements;
elastic auto-scale applications on Cloud resources based on their fluctuating load with optimised VM interoperability across Cloud infrastructures and without provider lock-in.
We gathered an interesting selection of complementary use cases from two SMEs and one industrial partner on energy control and management, earth observation and Cloud orchestration. For example, the WeSave use case provided by a Cloud provider called Wellness Telecom, is an application for energy control and optimisation of buildings that must be elastic to collect streams of data from various locations. WeSave is composed of various software components that require highly dynamic deployment and migration of VMs from one geographical location to another, as well as frequent starting and shutting down the VMs within seconds to provide satisfactory QoS to its end-users, a technology which is currently unavailable.
Small & medium enterprises, Open Source developers, Large companies, Technology providers, Research institutions.
Open Source software available on
The VM images will be evaluated for size, functionality, and delivery time, where the optimised images should be more than 60% than regular user-created VM images (comprising application users, OS experts and Cloud system administrators) while keeping their original functionality. We also expect to reduce the delivery time of the images from minutes to around 10 seconds for most cases. Even for more complex applications, where the original VM image delivery time is over 2 minutes, we will automatically create images with and reduce delivery time by more than 30%. For a given large set of VM images (at least 100) in a repository with over 100 GB of cumulative VM image size, we expect to reduce the storage requirements by more than 80%.
We also expect to reduce the time for cross data center deployment (i.e. when a VM is requested in a data center where the image is not available) by over 20\% compared to manual or semi-automatic techniques with no storage optimisation involved. The optimisation process should improve the performance of the pilot use cases by at least 30%. Moreover, the multi-objective approach should preserve performance while decreasing the costs and storage requirements by at least 25% compared to solutions without repository optimisation. The new VM management methods will improve the QoS elasticity of the use cases from their currently inelastic status to elastic. This will mean that the percentage of QoS change will be at least equal to, or exceed the percentage of change in resource provisioning.
Finally, the knowledge model will address interoperability and integration issues of the use cases, and will achieve an over 25% of productivity increase in their VM image preparation and deployment time.