Tuesday, March 15, 2011

Cloud computing for eScience users

As researcher working on tools and frameworks for porting scientific applications on distributed infrastructures, I found the second session of the first day of Cloudscape III here in Brussels very interesting around benefits and barriers of Cloud Computing for scientific research communities.

The session, chaired by Steven Newhouse, director of EGI, was introduced by a presentation by Guy Coates, Wellcome Trust Sanger Institute and member of the SIENA REB, on a use case for the biology and biotechnology community on the adoption of Cloud Computing. The benefits of Clouds for these communities come from IaaS solutions considered a useful building block for providing world-wide availability and reach of their services; also the provision of virtual hardware in remote cloud IaaS facilities from any location with internet access has several advantages on using internal hardware. On the SasS level, there is a vast eco-system of software that biology researchers typically want to run on their data; using a cloud model, researchers can upload data to the SaaS providers, who will then analyse, track and store their results. Researchers are therefore freed from having to manage their own HPC infrastructure.

The following talk from Ake Edlund, KTH Royal Institute of Technology, and member of the SIENA REB focussed on the activities of the Enabling Clouds for eScience (ECEE) group whose aim is to share experiences amongst European Cloud projects in order to find out as much as possible, how clouds can help users in their daily work.

The last talk from Alan Sill Open Grid Forum (OGF) and member of the SIENA REB, analyzed the importance of interoperability in Clouds, presenting the latest activities of the Open Cloud Computing Interface (OCCI) working group of OGF. OCCI specification is widely adopted in the cloud computing community and has attracted considerable interest from the commercial community and from other standards organizations due to its built-in inter-compatibility with other RESTful methods.

The concluding panel was joined by Ad Emmen, Genias Benelux and EDGI; Morris Riedel, Jülich Supercomputing Centre, EMI and member of SIENA REB; Alexander Papaspyrou, Technische University Dortmund and IGE. Benefits of cloud computing seem to overtake barriers for adoption; many scientific communities are perfectly suitable for moving to cloud environments: non HPC user communities for example submit small jobs thus being a valid candidate for shifting to clouds; climate changes communities applications are usually characterized by simple workflows easily portable to clouds. But several barriers are still blocking the shift, the biggest one being the performance on data that is preventing users to move data around due to networking lacks and interoperability issues.

The challenge is to benefit from lesson learned from Grid experience, designing smart environments that are "useful" for the researchers providing them not with complex tools but with familiar interfaces. In this direction has a wide consensus the need for the development of SaaS and PaaS solutions that could hide the complexity of the management of the infrastructure, leaving to experts the task of configuring resources and meeting the needs of the users for the easy porting of scientific applications to Clouds.

No comments: