Abstract
The CERN IT provides a set of Hadoop clusters featuring more than 5 PBytes of raw storage with different open-source, user-level tools available for analytical purposes. The CMS experiment started collecting a large set of computing meta-data, e.g. dataset, file access logs, since 2015. These records represent a valuable, yet scarcely investigated, set of information that needs to be cleaned, categorized and analyzed. CMS can use this information to discover useful patterns and enhance the overall efficiency of the distributed data, improve CPU and site utilization as well as tasks completion time. Here we present evaluation of Apache Spark platform for CMS needs. We discuss two main use-cases CMS analytics and ML studies where efficient process billions of records stored on HDFS plays an important role. We demonstrate that both Scala and Python (PySpark) APIs can be successfully used to execute extremely I/O intensive queries and provide valuable data insight from collected meta-data.
Highlights
The need for agile data analytics platforms is increasing in High Energy Physics (HEP)
To avoid these types of pitfalls we developed our own framework called CMSSpark [13] which provided all necessary tools for code submission to Spark platform, set up necessary Java libraries for data processing as well as set up proper run parameters optimized for CERN Spark cluster
We found that processing time to join and aggregate data across Data Bookkeeping System (DBS)/PhEDEx DBs and data-streams is impossible to achieve in a reasonable amount of time using a vertical approaches based on RDBMS solutions2
Summary
The need for agile data analytics platforms is increasing in High Energy Physics (HEP). Users are advised to rely on PySpark functions to avoid slowness of built-in Python functions over DataFrame operations, e.g. pyspark.sql.f unctions.split should be used instead of split Python built-in function To avoid these types of pitfalls we developed our own framework called CMSSpark [13] which provided all necessary tools for code submission to Spark platform, set up necessary Java libraries for data processing (e.g. support for CSV or Avro data formats) as well as set up proper run parameters optimized for CERN Spark cluster. Any user with access to the CERN Hadoop cluster can quickly launch distributed (Py)Spark job to process CMS data, focusing on what operation to run rather than struggling on how to properly submit YARN-based batch job. While looking at the current status of resource utilization, which is a very important component of our daily operations, we oversee that it can be further enhanced by predicting user behavior and adjusting our resources
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
