Wednesday, September 21, 2011

Lyon Rendez-vous

Suddenly we are starting day three of a very packed conference here in sunny Lyon. The EGI Technical Forum has so far proved to be a a very rewarding experience with a rich mix of participants in all of the sessions that I have attended. This has resulted in some great discussions and informative presentations. From my point of view as Chief Community Officer at EGI, I want to hear how the various research communities are getting on, what services and offerings are available for the communities and how efficiently all of this is working.

The three session Data Workshop which we ran on Monday (led by Gergely Sipos) proved to be particularly rewarding. The trigger for this and the rest of these User Support workshops is the ongoing analysis that EGI has been conducting on emerging requirements. The Data Workshop comprised three sessions; the first two were packed with a tight sequence of position papers where a whole range of experts presented their work relating to data. The first batch of these were from the EMI project, but many others contributed too including Steve Tuecke talking about Globus Online. The day ended with a fruitful discussion about the challenges relating to data in the light of the earlier presentations.

One theme that emerged was the "dissemination gap" that exists between existing technology and user needs. For a number of the declared requirements some claimed that solutions already exist. However, if solutions cannot be found and requirements are not promoted then we will not make these connections. Furthermore, software must be documented and easily available for this to happen. There was support for providing, or at least making available, more white papers about implementation as well as other formal comparison information on the services. But there was also discussion about the role that NGIs must play in distributing this sort of information to their local users.

There was another discussion about definitions and knowledge. We need to be much stricter and more consistent in how we categorise software and services. On example given was files - these are assumed to be persistent by some and transient by others. Context is everything, but if we cannot understand this then APIs alone won't solve the problems.

No comments: