Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
T: Fachverband Teilchenphysik
T 59: Grid-Computing und Software
T 59.6: Vortrag
Mittwoch, 27. März 2019, 17:15–17:30, S11
Integrating Dynafed into the ATLAS workflow — •Benjamin Rottler1, Frank Berghaus2, Felix Bührer1, and Markus Schumacher1 — 1Albert-Ludwigs-Universität Freiburg — 2University of Victoria
In the current ATLAS grid computing model analysis jobs are sent to the grid site where the corresponding data is available. With the ever increasing amout of measured and simulated data this is not feasible anymore in the future. Furthermore, there are plans that smaller grid sides can only provide computing ressources, but not storage.
Dynafed allows to aggregate different storage endpoints where data access is possible via the WebDAV protocol. It supports traditional grid storage solutions like dCache and DPM as well as object stores such as Amazon S3 or Microsoft Azure. Dynafed allows to redirect the user request to the nearest storage endpoint where the data is available.
Our goal is to incorporate Dynafed into the ATLAS grid workflow. For this we need to evalute the performance of Dynafed and WebDAV based access. We present results of those benchmarks and compare Dynafed with traditional access methods. Additionally, we compare the performance of grid storage and object store endoints.