220 likes | 317 Views
Experience of introducing e-infrastructure for laser performance analysis. Lakshmi Sastry, Srikanth Nagella, Richard Henwood, Victoria Marshall and Brian Matthews. Introduction to CLF. Target Area. Users visiting CLF bring their own diagnostics.
E N D
Experience of introducing e-infrastructure for laser performance analysis Lakshmi Sastry, Srikanth Nagella, Richard Henwood, Victoria Marshall and Brian Matthews
Target Area Users visiting CLF bring their own diagnostics. They still need to exchange information with the CLF team: Request a ‘shot’ of a given power Discover what CLF delivered
CLF workflow The laser CLF staff Users
CLF/e-Science Impact CLF: improved workflow:-> less effort delivering service
CLF/e-Science Impact CLF: improved workflow: -> less effort delivering service -> more effort to do Science -> more Nature covers!
CLF/e-Science Impact CLF: improved workflow: -> less effort delivering service -> more effort to do Science -> more Nature covers! e-Science: CLF data service:-> enabling more CLF Science
CLF/e-Science Impact CLF: improved workflow: -> less effort delivering service -> more effort to do Science -> more Nature covers! e-Science: CLF data service:-> enabling more CLF Science but also...-> eSc add value to User data-> more Nature covers!
CLF workflow The laser CLF staff Users
First, add value for CLF Functional requirements • 1000 diagnostic channels • Data bursts 20 seconds apart • 3 data types: 0d, 1d, 2d • Virtual channels • NeXus data format! • Stand-alone client • Flexible addition of new channels • Analysis capabilities of the client • Offsite access • Secure data collection for target area • Generalise to Vulcan
The Good • Pipeline handles real data bursts • Stand-alone client provides advanced analysis
The bad • Quality of service • No monitoring • No-one wants to Sys-admin • Quality of protection • Host and firewall are the same machine • No access control • Scalability • Increase burst rate?
DB Schema issues • Adding new channels requires a new _column_ on our DB table! • Eg Tracetable <50% useful data
DB Schema issues • Adding new channels requires a new _column_ on our DB table • Eg Tracetable <50% useful data • The Ugly
Managing risks Q: Will HTTP be fast enough? A: Project Scott
Instrumenting for e-Science Beyond thumb drive Sharing Pre-Conditions • Compelling service offering • CLF are enthusiastic • Funders data policy ?
Instrumenting for e-Science: Sharing data • The carrot • Governance
Instrumenting for e-Science: Sharing data • The carrot • Governance More carrots • Convenience • Social data • Maximising value of your work
Conclusions • Unique problem domain • MVC and SOA approach in phase 2 • Detailed benchmarking phase • Data governance is an important value-add • Great opportunity to do novel e-Science