Speaker
Quentin Guilloteau
Description
The development and evaluation of grid or cluster middlewares, such as batch schedulers, require to deploy numerous machines to reach an environment close to the full scale of the production system.
To avoid these huge deployments, one can consider folding the system on itself by deploying several "virtual" resources onto one physical resource.
In this study, we investigate the variations in performance for a distributed IO benchmark while folding a cluster of machines which contains a distributed file system.
This work is joint with Eric Rutten (INRIA) and Olivier Richard (UGA)
JLESC topic | experiments |
---|