BonFIRE-VCOC - BONFIRE-VIRTUAL CLUSTERS ON FEDERATED CLOUD SITES - Cesga

Project Description

BonFIRE-VCOC – BONFIRE-VIRTUAL CLUSTERS ON FEDERATED CLOUD SITES

Period: 2011-09-01 – 2012-11-30 (15 months).

Funded by: Comisión Europea

http://www.bonfire-project.eu

VCOC, Virtual Clusters on federated sitesis one out four new experiments in BonFIRE project which have been selected in the first open call. The aim of this experiment is to research the feasibility of using multiple Cloud environments for the provision of Services which need the allocation of a large pool of CPUs or virtual machines to a single user (as High Throughput Computing or High Performance Computing). It will use as use case the dose computation for radiotherapy treatments based on Monte Carlo methods, developed in the eIMRT project.

The experiment is looking for answers to 6 questions related to the usage of virtual clusters in a distributed Cloud environment. A first set of questions are related to the time that the deployment and enlargement of such cluster need to be operational and the influence of other simultaneously operations have in the process. Previously in other experiences in Grid, local clusters and even commercial Cloud providers, interferences with other users and processes have been observed which affect the final quality of service. The objective is to understand better how to manage these virtual clusters to guarantee that the time to solution or latency (this means, the time since the cluster has been requested to the end of the service). Also, the elasticity will be analyzed, using the application performance monitoring as trigger for the change in the size of the cluster. A second set of experiments will investigate the usage of the distributed capability of Cloud providers (federated or multi-site) in order to protect the service against failures (resilence). A virtual cluster will be deployed divided in two sets, and the characteristics of the network (latency, bandwidth and packet loss rate) will be changed to study the effects in the performance of the cluster. The radical situation of losing part of the cluster will be simulated changing the bandwidth to zero, when the other part should recover from it to guarantee that the customer receives the solution on time. Totally, nine experiments have been designed. Each execution will run a verification workflow of a radiotherapy treatment which simulates it using Monte Carlo techniques. The full process comprises five phases, two of them running hundreds of small jobs.

Publications:

[1] R. Valin, L. M. Carril, J. C. Mouriño, C. Cotelo, and C. Fernández, “Enabling the Deployment of Virtual Clusters on the VCOC Experiment of the BonFIRE Federated Cloud,” in CLOUD COMPUTING 2012, The Third International Conference on Cloud Computing, GRIDs, and Virtualization, 2012, no. c, pp. 237–242.