CMS Online Data Quality Monitoring: Real-Time Event Processing Infrastructure

The Online Data Quality Monitoring (DQM) system of the CMS experiment at the LHC processes CMS events in real time to produce histograms that are published on a DQM webserver for experiment-wide access. The CMS events are built, reconstructed and filtered by the CMS Data Acquisition System (DAQ) and the Higher Level Trigger (HLT) before they are written to disk using so-called StorageManager (SM) applications. The SM comprise event server queues that provide selected events to DQM consumer applications. For each CMS detector subsystem one or several consumer applications, implemented in the standard offline CMS software framework, are operated. Changing detector and machine conditions and improved understanding of quality criteria require the system to be flexible, simple and safe to update while being robust and fault tolerant at the same time. An overview of the design, implementation and operation of the DQM event server and processing infrastructure is given, detailing how the software components are controlled using the central CMS run control system and how they are interfaced with the standard offline CMS software. We review also how the system was improved recently, based on experience of operation in 2009, and leveraging multi-core hardware for improved performance in event processing