Description
Data management is becoming one of the key issues for eInfrastructures. Not only are scientific instruments and simulations producing increasing amounts of data, this data needs to be made available to worldwide collaborators in an efficient, reliable, and secure manner. In this session we discuss several aspects of data management on eInfrastructures, from providing reliable access to storage, efficient data transfers between web services, to data centric applications and their workflows.
-
Riccardo Zappi (INFN-CNAF)15/04/2010, 09:00Software services exploiting and/or extending grid middleware (gLite, ARC, UNICORE etc)OralStoRM is a high performance and flexible SRM solution for general disk based storage resources that bring in Grid the advantages of high performance storage systems based on cluster file system, such as GPFS from IBM and Lustre from SUN. StoRM can be used in small data centre with low rates of storage management requests and, at the same time, it is capable to grow in terms of storage and...Go to contribution page
-
Alvaro Fernandez Casani (Instituto de Fisica Corpuscular (IFIC) UV-CSIC)15/04/2010, 09:20Software services exploiting and/or extending grid middleware (gLite, ARC, UNICORE etc)OralServices needed to access scientific data and computational resources in a distributed computing infrastructure are getting more and more complex. Our Atlas Tier2/Tier3 users, and National e-Science and GRID-CSIC initiatives researchers, have different requirements but all need ubiquitous and efficient access to these resources. We will provide details on how we use WLCG and Glite middleware,...Go to contribution page
-
Mr Spiros Koulouzis (Informatics Institute, University of Amsterdam, the Netherlands)15/04/2010, 09:40Software services exploiting and/or extending grid middleware (gLite, ARC, UNICORE etc)OralDespite numerous benefits, many Web Services (WS) face problems with respect to data transport, either because SOAP doesn't offer a scalable way of transporting large data-sets or because orchestration workflows (WF) don’t move data around efficiently. In this paper we address both problems with the development of the ProxyWS. This is a WS utilizing protocols offered by the Virtual Resource...Go to contribution page
-
Mr Jakub MOSCICKI (CERN)15/04/2010, 10:00Scientific results obtained using distributed computing technologiesOralIn this work, we present an evaluation of a late-binding scheduler for HEP distributed analysis, where data is distributed globally with multiple replicas in a non-uniform way. We employ the late-binding technique in order to optimize job placements thereby minimizing the per-job time-to-completion. We evaluate different scheduling and prioritization strategies, and evaluate the approach...Go to contribution page