eMinerals project

Environment from the Molecular Level

A NERC eScience testbed project

Collaborative grids

A key role of escience is to enable scientists to work together in closer collaboration. A phrase that is often used to describe IT-enabled collaborations is the "Virtual Organisation" (VO). There are many different ideas of how a VO should work. Our own idea is that the eMinerals VO will work towards a vision in which scientists in different locations can work together as if they were in adjacent offices in one laboratory. To meet this end, we have a clear focus on establishing a collaborative grid with appropriate tools, together with a corresponding collaborative focus within our compute and data grids.

Collaborative grid

The eMinerals project has put some effort into exploring the tools that make the best contributions to the communication part of collaboration. Briefly we make use of the following tools

Desktop Access Grid. All team members are capable of running the Access Grid from their desktop or have an Access Grid suite very close at hand. We use the Access Grid for person-to-person, task-oriented and whole-project meetings. For example, the eMinerals project holds a weekly plenary meeting on the Access Grid. We have developed a multicast application sharing called (called MAST) to enhance the capabilities of our Access Grid meetings. One simple application for MAST is to run fortnightly seminars using powerpoint as a shared application. Face-to-face meetings are an essential part of collaboration: the communication gains of seeing the people you are talking to are sometimes invaluable. We find that a lot of work can be achieved during team Access Grid meetings.

Instant Messaging. Instant messaging (IM) enables partners to hold quick conversations with low-start up effort. Currently the eMinerals team uses AIM and iChat tools. We find that IM is ideal for cases when one partner needs to ask a quick question of another partner. In our experience, IM can be relatively non-intrusive for people capable of multi-tasking, and the productivity gains are significant. IM is certainly much more productive for communication than email.

Wiki. eMinerals runs its own wiki (with access enabled by the use of eScience X.509 digital certificates). The role of the wiki is to collate information from many sources, eg in planning a specific area of activity or in collating task lists. For example, the eMinerals project uses the wiki for shared writing of papers, collating information for reports, or in scoping smaller-scale collaborative projects.

Compute grid

Our work on collaborative computing is focussed on the eMinerals minigrid. Aside from the usual gains of using grid technologies to share compute resources, the eMinerals project fosters close collaboration between team members (both scientists and grid specialists) to maintain the minigrid infrastructure, develop new tools with in-built usability, to provide training in tools, and to provide trouble-shooting or answer questions.

Data grid and information delivery

A key component of any collaboration is to be able to share data. Although the eMinerals project is primarily known as a computational project, with an emphasis on developing compute grid infrastructures and tools, we have put a lot of effort into data management with a view to enabling sharing of data. Grid computing inevitably leads to a welcome deluge of data, and the challenge is to make data available to partner projects in a form that doesn't require collaborating scientists to understand the details of the simulation code used to generate the data. We call this "information delivery": making the information content of data available to collaborating scientists.

Our approach to data sharing within the eMinerals VO is to use the Storage Resource Broker (SRB) to store data. All project members have easy access to the SRB through a variety of client tools, including some developed within the eMinerals project. We have closely integrated the SRB within the eMinerals minigrid, and our job submission tools all use the SRB for data management.

Recent work has focussed on the use of metadata, including automatic capture of metadata as part of the job submission and management process. One important motivation for effective metadata creation and management is to enable collaborators find files based on content rather than via a file structure and hierarchy.

General references

Papers that describe the eMinerals collaborative grid work are:

"Collaborative tools in support of the eMinerals Virtual Organisation." MT Dove, M Calleja, R Bruin, J Wakelin, M Keegan, S Ballard, G Lewis, SM Hasan, V Alexandrov, RP Tyer, I Todorov, P Wilson, M Alfredsson, GD Price, C Chapman, W Emmerich, S Wells, A Marmier, S Parker, Z Du. Proceedings of the UK e-Science All Hands Meeting 2004, (ISBN 1-904425-21-6), pp 127–134, 2004

"The eMinerals collaboratory: tools and experience." MT Dove, M Calleja, R Bruin, J Wakelin, MG Tucker, GJ Lewis, SM Hasan, VN Alexandrov, M Keegan, S Ballard, RP Tyer, I Todorov, PB Wilson, M Alfredsson, GD Price, C Chapman, W Emmerich, SA Wells, A Marmier, SC Parker, Zhimei Du. Molecular Simulation 31, 329–337, 2005

"The eMinerals project: developing the concept of the virtual organisation to support collaborative work on molecular-scale environmental simulations." MT Dove, E Artacho, TO White, RP Bruin, MG Tucker, P Murray-Rust, RJ Allan, K Kleese van Dam, W Smith, RP Tyer, I Todorov, W Emmerich, C Chapman, SC Parker, A Marmier, V Alexandrov, GJ Lewis, SM Hasan, A Thandavan, K Wright, CRA Catlow, M Blanchard, NH de Leeuw, Z Du, GD Price, J Brodholt, M Alfredsson. Proceedings of All Hands 2005 (ISBN 1-904425-53-4), pp 1058–1065, 2005