From kurca@in2p3.fr Wed Oct 11 11:02:49 2006 Date: Tue, 10 Oct 2006 19:59:39 +0200 (CEST) From: Tibor Kurca To: osg-council@opensciencegrid.org Cc: Ruth Pordes , Brad Abbott Subject: Acces to OSG resources Dear Sir, Madam, In June of this year the D0 collaboration successfully reestablished its data taking after the start of Tevatron Run IIb. The amount of data is growing very quickly and the computing resources of D0 are limited. We have to assure the primary processing of newly taken data and at the same time, after the new calorimeter calibration has been completed, to reprocess the data from Run IIb taken up to now. As you know, D0 has been successfully promoting grid computing for a long time. We are using computing resources at D0 institutes distributed all around the world via SAMGrid. We have also pushed the concept of grids interoperability and are already using the LCG resources via SAMGrid/LCG interface for our Monte Carlo production. We are working hard together with OSG experts on the SAMGrid/OSG interoperability project. We have reached the phase where we are able to submit SAMGrid jobs for execution on the OSG clusters. We will need to reprocess approximately 500 million events on the timescale of a few months. The deadline for finishing this task is March 2007, when the data have to be ready for physics analysis. To be able to accomplish this enormous task we would like to kindly ask for access to your computing resources. We hope that all code is in place except for the final calibrations by the beginning of November. We would like to perform a pre-production run starting in early November to ensure all sites are operating smoothly and to solve any possible problems. We then would like to start full production running on December 1 2006. We estimate that we would need about 1500-2000 CPUs for a period of about 4 months. We would like to identify the OSG sites as soon as possible to allow us to begin to set up the required DZero specific servers (SAM-stations) , identify the bandwidth available, identify the necessary manpower to send jobs,etc. Some detailed information about the resources required are listed below: For this reprocessing task we would like to use the OSG Resource Selection Service (ReSS) therefore sites that will run DZero jobs are encouraged to install the local services. This would help a lot our job and workload management. For SAMGrid jobs to run we require: 1. Worker nodes need to have outgoing network access. This is an absolute must since we contact DZero data management services (SAM) at FNAL, SAM-stations, SAM disks and durable storages which will be outside the OSG site, as well as calibration servers at FNAL, etc. 2. Worker nodes should have about 6GB of scratch storage space per job. Disk Requirements: We would need about 4 TB of total disk space to store input and output files. This space can be distributed among more participating sites. For prestaging of our raw data it would be desirable to have at least 1 TB of disk cache connected with 1 Gbit link to all CPUs. This storage space will be managed by SAM, therefore we would like to have them installed on sites with active Dzero collaborators. This should help us with operation and maintenance. We would appreciate very much your help. We believe that the successful completion of this task would be a great achievement for the whole grid community and it would be a proof of the viability of grids interoperability projects. best regards Tibor Kurca D0 reprocessing coordinator +------------------------------------------------------------------+ | Tibor Kurca email: kurca@in2p3.fr | | Centre de Calcul-IN2P3-CNRS Tel: +33 (4)-72-69-42-02 | | 27, Bd. du 11 Novembre 1918 Fax: +33 (4)-72-69-41-70 | | Bat 12. 12-14 Bd. NIELS BOHR | | F-69622 Villeurbanne, Cedex | | | | Institute de Physique Nucleaire de Lyon, Groupe D0 | | 43, Bvd du 11 Novembre 1918 Tel: +33 (4)-72-44-85-01 | | F-69622 Villeurbanne, Cedex Fax: +33 (4)-72-43-14-52 | +------------------------------------------------------------------+