An SQL-based approach to physics analysis

20th International Conference on Computing in High Energy and Nuclear Physics

物理学;计算机科学

Limper, Dr Maaike^1

CERN IT Department, Geneva 23

CH-1211, Switzerland^1

关键词: Complex analysis;    Large Hadron collider LHC;    LHC computing grids;    Physics analysis;    Read speed;    Relational Database;    SQL query;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/513/2/022022/pdf
DOI  :  10.1088/1742-6596/513/2/022022

学科分类：计算机科学（综合）

来源: IOP

【 摘 要 】

As part of the CERN openlab collaboration a study was made into the possibility of performing analysis of the data collected by the experiments at the Large Hadron Collider (LHC) through SQL-queries on data stored in a relational database. Currently LHC physics analysis is done using data stored in centrally produced "ROOT-ntuple" files that are distributed through the LHC computing grid. The SQL-based approach to LHC physics analysis presented in this paper allows calculations in the analysis to be done at the database and can make use of the database's in-built parallelism features. Using this approach it was possible to reproduce results for several physics analysis benchmarks. The study shows the capability of the database to handle complex analysis tasks but also illustrates the limits of using row-based storage for storing physics analysis data, as performance was limited by the I/O read speed of the system.