Current Grid operation and future role of the Grid
O. Smirnova · Journal of Physics Conference Series · 2012
Grid-like technologies and approaches became an integral part of HEP experiments.Some other scientific communities also use similar technologies for data-intensive computations.The distinct feature of Grid computing is the ability to federate heterogeneous resources of different ownership into a seamless infrastructure, accessible via a single log-on.Like other infrastructures of similar nature, Grid functioning requires not only technologically sound basis, but also reliable operation procedures, monitoring and accounting.The two aspects, technological and operational, are closely related: weaker is the technology, more burden is on operations, and other way around.As of today, Grid technologies are still evolving: at CERN alone, every LHC experiment uses an own Grid-like system.This inevitably creates a heavy load on operations.Infrastructure maintenance, monitoring and incident response are done on several levels, from local system administrators to large international organisations, involving massive human effort worldwide.The necessity to commit substantial resources is one of the obstacles faced by smaller research communities when moving computing to the Grid.Moreover, most current Grid solutions were developed under significant influence of HEP use cases, and thus need additional effort to adapt them to other applications.Reluctance of many non-HEP researchers to use Grid negatively affects the outlook for national Grid organisations, which strive to provide multi-science services.We started from the situation where Grid organisations were fused with HEP laboratories and national HEP research programmes; we hope to move towards the world where Grid will ultimately reach the status of generic public computing and storage service provider and permanent national and international Grid infrastructures will be established.How far will we be able to advance along this path, depends on us.If no standardisation and convergence efforts will take place, Grid will become limited to HEP; if however the current multitude of Grid-like systems will converge to a generic, modular and extensible solution, Grid will become true to its name.