3 Mass Data Processing Optimization on High Energy Physics Experiments
Download citation file:
- Ris (Zotero)
- Reference Manager
Experiments in the field of particle physics produce vast amounts of data which need to be stored and processed. Reliable and efficient data acquisition system is therefore an integral part of these experiments. This article presents the database optimization work that has been done on the COMPASS experiment operating on the Super Proton Synchrotron particle accelerator in CERN (European Organization for Nuclear Research) in years 2010 and 2011. Authors present the complex optimization methodology step by step, with the belief that such procedure can and should be successfully re-executed on any other complex database system, where enhancing speed and reliability of the data management becomes a task of extreme importance. The database service of the COMPASS experiment suffered from the overload during the 2009 run, thus it has been decided to update the entire database system in all aspects, from hardware improvements through new architecture and software upgrades to storage engine optimization, reviewing table structures and query performance tuning. In this paper, the proposal and the implementation of these new improvements is presented. The updated architecture employs replication, continuous monitoring, and regular backups to achieve high reliability and availability of the database service. Improvement process is described in detail and the performance of the new system is evaluated.