Speaker
Description
Authors: Maria-Mateea Popescu (CERN, maria.mateea.popescu@cern.ch),
Costin Grigoraș (CERN, costin.grigoras@cern.ch),
Cristian Mărgineanu (National University of Science and Technology Politehnica Bucharest, cristian.margineanu@stud.acs.upb.ro)
on behalf of the ALICE collaboration
MonALISA serves as the monitoring backbone for the distributed computing infrastructure of the ALICE experiment at CERN, currently managing approximately 17 million parameters published by all components of the Grid middleware, with a global update rate of around 200 kHz. The monitoring framework also implements distributed data aggregation, allowing clients to subscribe to high-level metrics of the system as well as offering real-time access to each parameter for debugging purposes using different types of visualisation. While the system was robust and served the ALICE experiment well over the last two decades, its reliance on the obsolete Jini (Apache River) framework and the underlying Java RMI framework, now deprecated, creates a critical maintainability issue.
In this paper, we will present a comprehensive modernisation of MonALISA, transitioning away from the obsolete technologies to an open architecture that is made to last for the next decades. The updated monitoring components rely on WebSockets for real-time communication and streaming of data, making the entire framework language-independent for easier interoperability with external components. Furthermore, we describe a complete replacement of the user interface layer, originally built using Java AWT and Java Web Start, with a modern, lightweight web-based application. We will also present the migration plans, having designed the new system to be compatible with the old one for the transition period, thus allowing the distributed services as well as long running clients (like the central data collectors of the experiment) to be upgraded without impacting the operations.