The Monitoring Data Archiving Service for ATLAS

ATLAS is one the four experiments being assembled at the CERN Large Hadron Collider (LHC). The complexity of the ATLAS experiment and the high event rate make the monitoring system an essential tool to assess the status of the hardware and the quality of the data while they are being acquired. It is important that all the monitoring data, mainly ROOT histograms, are saved to a permanent storage system, so that they can be used later for studying the time evolution of the experimental conditions or as reference for the future runs. The presentation will show the solution proposed to this issue within the ATLAS Trigger and Data Acquisition (TDAQ) project. Many GB of monitoring data are expected per run. At the end of each run, the Monitoring Data Archiving service (MDA) retrieves all the available histograms from the Online Histogramming service (a temporary storage provided within the ATLAS TDAQ software framework) and writes them into ROOT files, which in turn will be stored on tapes. The Collection and Cache service (CoCa) manages a disk based cache in order to guarantee a fast access to the histograms produced during the last runs. Furthermore, it collects many small files and produces big archives to be stored on tape, thus enhancing the efficiency in the tape usage. Initially meant as a component of MDA, CoCa has evolved as an independent package that can be used by any ATLAS online application.