ICALEPCS2023 - List of Authors (Baggiolini, V.)

Paper	Title	Page
THPDP068	Implementing High Performance & Highly Reliable Time Series Acquisition Software for the CERN-Wide Accelerator Data Logging Service	1494
	M. Sobieszek, V. Baggiolini, R. Mucha, C. Roderick, P. Sowinski, J.P. Wozniak CERN, Meyrin, Switzerland
	The CERN Accelerator Data Logging Service (NXCALS) stores data generated by the accelerator infrastructure and beam related devices. This amounts to 3.5TB of data per day, coming from more than 2.5 million signals from heterogeneous systems at various frequencies. Around 85% of this data is transmitted through the Controls Middleware (CMW) infrastructure. To reliably gather such volumes of data, the acquisition system must be highly available, resilient and robust. It also has to be highly efficient and easily scalable, given the regularly growing data rates and volumes, particularly for the increases expected to be produced by the future High Luminosity LHC. This paper describes the NXCALS time series acquisition software, known as Data Sources. System architecture, design choices, and recovery solutions for various failure scenarios (e.g. network disruptions or cluster split-brain problems) will be covered. Technical implementation details will be discussed, covering the clustering of Akka Actors collecting data from tens of thousands of CMW devices and sharing the lessons learned. The NXCALS system has been operational since 2018 and has demonstrated the capability to fulfil all aforementioned characteristics, while also ensuring self-healing capabilities and no data losses during redeployments. The engineering challenge, architecture, lessons learned, and the implementation of this acquisition system are not CERN-specific and are therefore relevant to other institutes facing comparable challenges.
	Poster THPDP068 [2.960 MB]
DOI •	reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-THPDP068
About •	Received ※ 06 October 2023 — Revised ※ 12 October 2023 — Accepted ※ 13 October 2023 — Issued ※ 20 November 2023
Cite •	reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)

Paper

Title

Page

Implementing High Performance & Highly Reliable Time Series Acquisition Software for the CERN-Wide Accelerator Data Logging Service

1494

M. Sobieszek, V. Baggiolini, R. Mucha, C. Roderick, P. Sowinski, J.P. Wozniak
CERN, Meyrin, Switzerland

The CERN Accelerator Data Logging Service (NXCALS) stores data generated by the accelerator infrastructure and beam related devices. This amounts to 3.5TB of data per day, coming from more than 2.5 million signals from heterogeneous systems at various frequencies. Around 85% of this data is transmitted through the Controls Middleware (CMW) infrastructure. To reliably gather such volumes of data, the acquisition system must be highly available, resilient and robust. It also has to be highly efficient and easily scalable, given the regularly growing data rates and volumes, particularly for the increases expected to be produced by the future High Luminosity LHC. This paper describes the NXCALS time series acquisition software, known as Data Sources. System architecture, design choices, and recovery solutions for various failure scenarios (e.g. network disruptions or cluster split-brain problems) will be covered. Technical implementation details will be discussed, covering the clustering of Akka Actors collecting data from tens of thousands of CMW devices and sharing the lessons learned. The NXCALS system has been operational since 2018 and has demonstrated the capability to fulfil all aforementioned characteristics, while also ensuring self-healing capabilities and no data losses during redeployments. The engineering challenge, architecture, lessons learned, and the implementation of this acquisition system are not CERN-specific and are therefore relevant to other institutes facing comparable challenges.

Poster THPDP068 [2.960 MB]

DOI •

reference for this paper ※ doi:10.18429/JACoW-ICALEPCS2023-THPDP068

About •

Received ※ 06 October 2023 — Revised ※ 12 October 2023 — Accepted ※ 13 October 2023 — Issued ※ 20 November 2023

Cite •

reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)