Frédéric Desprez, Chief Senior Research Scientist, Inria
DIET, a scalable platform for clusters, grids and Clouds
In recent years, large-scale (and distributed) storage and computing have proven to be mandatory in IT. Internet Computing and Storage have considerably evolved from small isolated nodes to large-scale cluster-like architectures driven by efficiency and scalability, which we now know as "Clouds". Their goal is to be dynamically scalable and offer virtualized resources as a service over the Internet. Usually, solutions deployed in Clouds are aimed at web browsers and are load balanced when it comes to computational power and storage. Clouds can also be used in more computational-intensive domains as scalable computational resources. From a middleware standpoint, Cloud infrastructures introduce new sets of resources with different features. For this reason, middleware environments should be extended to manage these platforms.
Started in 2001, the DIET project is focused on the development of a scalable middleware, with initial efforts concentrated on distributing the scheduling problem across a hierarchy of agents. At the top of that hierarchy sits the Master Agent (MA), with Service Daemon agents at the leaf level. First focused on Grids, the project evolved to target Clouds platforms as well. Validated over Grid'5000, the French large-scale instrument for computer science, it has now been transfered to SysFera, an Inria spin-off for providing seamless access to clusters, grids and Clouds (in particular through a simple web portal). Using the DIET Cloud, it is possible to deploy a large-scale distributed and secure HPC platform that spans a large pool of resources, aggregated from different providers. Moreover, thanks to cutting-edge advances in research, the platform is made easier to use and can automatically adapt to the users' needs. Users will be able to submit large computational jobs as well as Big Data jobs, without losing control of their data. With almost no additional cost, they can take advantage of all the resources they can access.
In this talk, we will deal with the whole platform, its use over grids and Clouds, and our most recent work around Clouds and virtualized platforms.
Frédéric Desprez is a Chief Senior Research Scientist at Inria and holds a position at the LIP laboratory (ENS Lyon, France). He co-founded the SysFera company where he holds a position as scientific advisor. He received his PhD in C.S. from Institut National Polytechnique de Grenoble, France, in 1994 and his MS in C.S. from ENS Lyon in 1990.
His research interests include parallel algorithms, scheduling for large scale distributed platforms, data management, and grid and cloud computing. He leads the Grid'5000 project, which offers a platform to evaluate large scale algorithms, applications, and middleware systems.
See http://graal.ens-lyon.fr/~desprez/ for further information.