STAR Analysis Infrastructure. Tools for beginners
The STAR detector tracks thousands of particles (see Fig.1) produced in Au + Au, U + U, Cu + Cu and polarized p + p collisions. While STAR collects physics data, there are always large volumes of accompanying meta-data to track, store and analyze. The STAR detector is composed of over two dozen subsystems operating in concert while RHIC provides colliding beams. Hundreds of scientists and engineers are watching the data-taking process and tuning detector performance by checking meta-data streams produced by detector components. To ease the procedure of collection, analysis and review of that meta-data, STAR software group created an MQ-based MIRA framework.
Fig. 1. STAR U + U event at energy 193 GeV per nucleon
The STAR Experiment further exploits scalable message-oriented model principles to achieve a high level of control over online data streams. In this paper we present an AMQPpowered Message Interface and Reliable Architecture framework (MIRA), which allows STAR to orchestrate the activities of Meta-data Collection, Monitoring, Online QA and several Run-Time and Data Acquisition system components in a very efficient manner. The very nature of the reliable message bus suggests parallel usage of multiple independent storage mechanisms for our meta-data. We describe our experience with a robust data-taking setup employing MySQL- and HyperTable-based archivers for meta-data processing. In addition, MIRA has an AJAX-enabled web GUI, which allows real-time visualisation of online process flow and detector subsystem states, and doubles as a sophisticated alarm system when combined with complex event processing engines like Esper, Borealis or Cayuga. The performance data and our planned path forward are based on our experience during the 2011−2012 running of STAR.
Online Meta-data Collection and Monitoring Framework, Dmitry Arkhipkin, Jerom Lauret, Wayne Betts, Gene Van Burden (PDF)