Development of the Online Data Processing System for the BM@N Experiment at NICA

A huge amount of experimental data should be collected, stored and processed in large modern high-energy physics experiments, including the experiments of the NICA (Nuclotron-based Ion Collider fAcility) project at the Joint Institute for Nuclear Research. In this regard, corresponding performance requirements are put forward for existing online systems. The online data processing system developed for the BM@N experiment within the NICA project is based on a distributed architecture, enabling it to meet high performance requirements through scalability and parallel computing. The purpose of the online system is selective data processing (conversion to event digits in the CERN ROOT format and fast event reconstruction) and data monitoring of the ongoing experiment. To achieve this goal, the FairMQ package implemented by the FAIR collaboration (GSI Institute, Germany) has been chosen to communicate distributed processes executed on the nodes of the computing infrastructure with each other through the exchange of their messages. One of the issues in developing and using such systems is the problem of the distributed run and control of the processes. The task has been solved by using the FAIR DDS (Dynamic Deployment System) toolkit. The BM@N online system starts the predefined software tasks in a required sequence and allows managing them during sessions, including the transmission of messages between the tasks and the update of some properties. The paper presents the purposes and architecture of the Online Data Processing System for the BM@N experiment and features of the current implementation.