Go to Home Page
CHEP INFORMATION
Bulletins
Committees
Scientific Program
Docs by topics
Social events
Conference location
Secretariat
GRID INFORMATION
 • Grid WShop & Tutorial
 • Grid Program
USEFUL LINKS
 • Visiting Padova
 • INFN Padova
 • University of Padova
 • CHEP: '94 '95 '97 '98

Last update: Apr 1, 2000

to first abs  to previous absby abs number to next abs  to last abs

 

to first abs on this KT  to previous abs on this KTon same keytopic to next abs on this KT  to last abs on this KT


C078

New Data Storage Model for H1

Christoph Grab1, Ralf Gerhards2, Ursula Berthon3, Thomas Benisch2, Thomas Hadig4
  1. DESY/ ETH-Zuerich
  2. DESY
  3. Ecole Polytechnic Paris
  4. RWTH-Aachen

Speaker: Thomas Benisch

  The electron proton collider HERA at the DESY laboratory in Hamburg and the H1 experiment are performing substantial upgrades during the years 2000 and 2001. To cope with this expected increased luminosity and therefore the increased demands on data storage and data handling, the H1 Collaboration is pursuing a major revision of its computing environment. This context makes the optimization of data access an issue of extreme importance. It will be necessary to split the event data into several parts according to the frequency of their access and to keep these parts on different levels of a multilevel hierarchical storage system. We expect that a mode of data access in which only part of the event data is read will be essential for many physics analyses. We foresee a four tier storage model: the lowest level consists of the RAW data stored on tape, the second level consists of the so called ODS (object data system), that basically contains all standard reconstructed entities, such as tracks and clusters. The third tier is micro-ODS that contains high-level particle objects on the 4-vector level. The top tier finally is a tag event database, which allows dynamic event selection based on kinematical and topological event topologies.
   Recently, the H1 collaboration opted to use ROOT I/O for the data handling system after the lumi-upgrade. Important features of the ROOT I/O model which lead to this choice include support of both sequential and direct data access techniques, support of networking, the possibility of splitting event data into several streams and to write these streams into one or several files. ROOT allows the use of powerful object-oriented techniques for accessing the data.
   Tools provided by the ROOT project allow the H1 collaboration to make a smooth adiabatic transition from the present sequential event data access to a scheme based on direct data access, with the event data being split into several different files. H1 plans to make this transition in several steps.
   We present the general strategy taken by the H1 experiment and related implementation issues. We report on the results of prototyping ROOT-based I/O in H1 and performance benchmarks.

Presentation:  Adobe Acrobat pdf Short Paper:  Adobe Acrobat pdf 



  | Top | Home | Bulletins | Committees | Scientific Program | Docs by topics | Social Event | Conference Location | Secretariat | Privacy Policy |