The Tivoli Storage Manager in the Large Hardron Patrick Collider - PowerPoint PPT Presentation

dCache.ORG dCache.ORG The Tivoli Storage Manager in the Large Hardron Patrick Collider Grid world Fuhrmann for the dCache people TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG LCG Tier Center Mechanism Tier 2 Centers Disk only Regional Centers Aachen Tier 1 Centers Disk and Tape Japan IN2P3(Lyon),FR DESY Karlsruhe,GR Munich Tier 0 (CERN) 2.5 GBit/sec others *** 2.5 – 10 GBit/sec Italy e.t.c. Madrid,SP Tokio,JP Brookhaven,US FermiLab,US RAL,UK Canada Amsterdam, NL Spain US Netherlands TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG LCG Tier Center Mechanism (cont.) Job Workload Manager Resource Broker Job Submission LCG Site Compute Element LCG Site LCG Site Storage Element LCG Site Compute Element LCG Site Compute Element LCG Site Compute Element Storage Element Compute Element Storage Element Compute Element Data Flow Storage Element Storage Element Storage Element TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG Tier 0/1 Compute and Storage Element Setup Tier 1 Center A Compute Element (CE) Storage Element (SE) Local Access Prot. RFIO, dCap, xrootd Disk Storage Tape Storage Worker Nodes Remote Protocols Steering : File Transfer Service ( FTS ) Control : Storage Resource Manager ( SRM ) Transfer : GsiFTP / GridFTP Tier 1 Center B Compute Element (CE) Storage Element (SE) Local Access Prot. RFIO, dCap, xrootd Disk Storage Tape Storage Worker Nodes TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG Storage Element Requirements Being a LCG Storage Element, means support of local (posix like) access protocol : rfio, dCap, (xrootd) support of the GsiFTP data transfer protocol Secure Wide Area data transfer protocol support of the Storage Resource Manager Protocol ( SRM ) Checksum management Space allocation Transfer Protocol Negotiation Dataset pinning Third Party Transfer on behave of dataset owner support of the GRIS information provider protocol Availability Free/Available Space Access Protocol and access details (URIs) TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG The dCache Storage Element TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG The dCache Storage Element dCache : Managed Disk Storage System Basic Specification Single 'rooted' file system name space tree File system names space view available through an nfs2/3 interface Data is distributed among a huge amount of possible cheap disk servers. Supports multiple internal and external copies of a single file Supports 'posix like' (authenticated) access as well as various FTP dialects and the Storage Resource Manager Protocol. TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG dCache : Scalability Scalability Distributed Movers AND Access Points (Doors) Automatic load balancing using cost metric and inter pool transfers. Pool 2 Pool transfers on pool hot spot detection Handles bunch requests by fast pool selection unit TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG dCache : Configuration, HSM Configuration Fine grained configuration of pool attraction scheme. (write pools, subnet, directory tree, storage info) Pool to pool transfers on configuration of forbidden transfers Fine grained tuning : Space vs. Mover cost preference Tertiary Storage Manager connectivity Automatic HSM migration and restore HSM dCache interface by script (shell, perl ...) Convenient HSM connectivity for enstore, osm, TSM, Hpss TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG Resilient dCache Resilient dCache Controls number of copies for each dCache dataset Makes sure n < copies < m Adjusts replica count on pool failures Adjusts replica count on scheduled pool maintenance Embedded farm node dCache (makes use of local disk space) TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG dCache : Storage Element LCG Storage Element Local Access Protocol : dCap , Posix like access GsiFtp support SRM version ~ 1 (1.7) plus Space Reservation limited GRIS functionality, will improve soon TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG dCache development and support Topology dCache Collaboration San Diego Super Computer Center Virtual Organization Management Fnal DESY INFN Bari Information Provider CDF gridPP Deployment RAL US CMS Brookhaven LCG Deployment Jane ???? LCG Tier I / II UK CERN Grid KA Karlruhe LCG Tier I / II TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG dCache module Topology Grid Gris Layer Storage Resource Mgr (SRM) Access (gsi,kerberos) dCap Server Layer dCap Client Ftp Server (gsi, kerberos) http / https Extended Resilient Manager HSM Flush Manager Prestager HSM Flush Manager central services Core Layer dCache Core TSM Adapter Pnfs Cell Package TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG HSM interactions HSM Interactions TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG Dataset life cycle T S M d C a c h e C l i e n t precious cached S p a c e n e e d e d F i l e r e q u e s t e d cached TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG Deferred HSM flush Data collected per Storage Class (files space, subdirectory trees) Collected data flushed following rules (space,time,# of files) Tape Mount Data Flow Data Transferred Client -> dCache dCache -> HSM Time TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG dCache transfer scalability Scalability by highly distributed Storage Pools and direct client – disk and disk – hsm connection Client(s) http(s) GridFtp dCap No transfer funnels or bottlenecks TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG TSM, the ideal dCache tape backend The TSM is the ideal dCache tape storage backend In place at all big lab's (mostly for backup) Mature and well established software Drivers available for large set of tape libraries Administration knowledge : wide spread Already in place (as dCache back end) : Jülich : planned for this year Canada, TRIUMF GridKa, Karlruhe Sara, Amsterdam via SGIs DMF TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG Big Question Does TSM scale with the upcoming requirements ? And how ??? TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann

dCache.ORG dCache.ORG www.dCache.ORG TSM Symposium, Oxford Sep 27, 2005 Patrick Fuhrmann