90 likes | 195 Vues
This document outlines the implementation of distributed storage solutions through REDDnet data depots, leveraging the IBP protocol for managing CMS datasets at Vanderbilt University. Covering aspects such as remote job execution on Tier 2 and Tier 3 sites, the integration of gridFTP gateway for data ingestion, and essential CMS interfaces for file movement, this resource provides insights into current operations and future plans for enhanced data streamlining. With a deployment status highlighting 200 TB currently in place and future expansions, it's a crucial guide for T3 users interested in collaborative data handling.
E N D
REDDnet & CMS T3 Users Daniel Engh Vanderbilt University Oct 23, 2008
REDDnet distributed storage Distributed Data depots -- based on IBP protocol Tier 2, Tier 3, or OSG site CMSSW job Running remotely Subscribe to CMS datasets, injest via gridFTP gateway REDDnet depots PhEDEx / GridFTP gateway CMS Root Analysis running on local system CMS User
CMS Standard Interfaces • File Movement • Gridftp/SRM/Phedex • Gridftp/L (REDDnet DSI backend) • Usage: gsiftp://gridftp.reddnet.org/LFN • Event-data streaming • Root I/O • File/XRootD • Root/L (reddnet root-posix plugin) • Usage: TFile::Open(“lodn://lodn.reddnet.org/LFN”) • Security • Grid certificates/VOMS to write/manage data • Root read-only streaming access currently open
CMS Use Cases • Automatic Reddnet access • T3 site configured with REDDnet backend • Vanderbilt set up • Data registered with Phedex • Jobs submitted via CRAB • User Configured • Uploads data • Plans replication strategies • Configures jobs • Minimal local hardware requirements
Deployment status & plans • Depots: 200 TB @ 8 sites • 500TB additional planned • REDDnet T3 site • Vanderbilt • T3 test group • Caltech, UFL • General availability for interested T3 users • Jan 2009 • Future: • Async tools • More flexible data distribution