O2 (ALICE upgrade)

O2 (ALICE upgrade)
Name	O2 (ALICE upgrade)
Location	CERN, Geneva
Operated by	CERN

O2 (ALICE upgrade) is the integrated online-offline computing system developed for the ALICE experiment at CERN to handle the high-rate data from the Large Hadron Collider during Run 3 and beyond. It replaces the separate online trigger and offline reconstruction chains with a unified architecture that performs real-time data reduction and calibration to enable prompt physics analysis for heavy-ion collisions and proton-proton measurements. The system integrates custom hardware, distributed storage, and scalable software to meet the throughput demands of upgraded detectors such as the Time Projection Chamber and new inner tracking systems.

Overview

O2 was conceived to process the high-luminosity output of the Large Hadron Collider following the Long Shutdown 2 upgrades to the ALICE detector, addressing challenges from increased collision rates at the LHC and enhanced readout from the Time Projection Chamber, Inner Tracking System, and forward detectors. The project coordinated resources across CERN computing centers, national laboratories such as GSI Helmholtz Centre for Heavy Ion Research, Brookhaven National Laboratory, and institutes including Institut de Physique Nucléaire d'Orsay and University of Birmingham. O2 leverages commercial multi-core CPUs, general-purpose GPUs similar to deployments at Fermilab and Brookhaven National Laboratory, and network fabrics parallel to those used by EMBL and DESY for high-throughput scientific computing.

Motivation and Objectives

The primary motivation was to enable continuous readout and real-time processing for high-rate heavy-ion physics following detector upgrades completed during Long Shutdown 2. Objectives included replacing hardware triggers used in experiments like ATLAS and CMS with a flexible software-based approach inspired by initiatives at LHCb and NA62, reducing raw data volumes in a way comparable to compression strategies at IceCube and KM3NeT, and providing a unified framework for prompt calibration akin to systems at Belle II and BaBar. O2 sought to ensure compatibility with distributed computing models used by the Worldwide LHC Computing Grid, while facilitating interactive analysis workflows used by collaborations such as ALICE, ATLAS, and CMS.

Architecture and Components

The O2 architecture combines front-end readout electronics interfacing with ALICE subdetectors such as the Time Projection Chamber, Transition Radiation Detector, Muon Spectrometer, and Inner Tracking System. Data are sent to a farm of readout nodes and Event Processing Nodes inspired by computing farms at CERN IT and Brookhaven National Laboratory. Core components include the Common Readout Receiver Card (similar in role to custom FPGA solutions deployed by LHCb), a data transport layer modeled after middleware like that used at GSI and DESY, and a storage layer integrating local SSD caches and distributed file systems akin to EOS and Ceph used at CERN and EMBL. The processing stack uses heterogeneous compute resources with GPU acceleration comparable to deployments at NVIDIA-enabled clusters and CPU-based processing nodes as in Oak Ridge National Laboratory systems.

Data Processing and Software Framework

O2's software framework unifies online reconstruction, calibration, and quality assurance with offline analysis tools. It builds upon middleware and libraries used in ROOT-based analysis chains common across CERN experiments and integrates workflow orchestration similar to systems used by ATLAS PanDA and CMS CRAB. The framework adopts modular components for tracking, particle identification, and calibration developed by ALICE working groups together with contributions from institutes such as Universität Heidelberg, Czech Technical University, and Politecnico di Milano. Data formats and serialization follow patterns employed in HDF5-style scientific stacks and are compatible with distributed metadata catalogs like those used by WLCG partners including RAL and GridKA.

Performance and Commissioning

Performance validation involved stress tests and commissioning runs during the end-of-year technical stops and pilot beams at the LHC, coordinated with experiments including ATLAS and CMS to ensure machine compatibility. Benchmarks measured sustained throughput, latency, and compression ratios against targets derived from physics cases by ALICE analysis teams and external review committees with participants from BNL, GSI, and INFN. Commissioning used detector calibration sequences employed historically in ALICE and cross-checked with calibration techniques from LHCb and STAR to validate tracking efficiency, momentum resolution, and particle identification under realistic pileup conditions.

Physics Impact and Use Cases

By enabling continuous real-time reconstruction and calibration, O2 allows ALICE to deliver high-statistics measurements of quark–gluon plasma observables, heavy-flavor production, and rare probes studied in analyses comparable to those by PHENIX, STAR, and CMS. Use cases include precision flow measurements, jet quenching studies, and quarkonia suppression analyses that require prompt access to reconstructed events comparable to workflows at ATLAS and LHCb. The system supports fast-turnaround physics production for working groups within the ALICE collaboration and facilitates multi-institution efforts spanning CERN, INFN, CNRS, and national labs such as BNL and GSI.

Timeline and Collaboration

Development started after design reviews preceding Long Shutdown 2, with hardware procurement and firmware development coordinated with vendors and partner laboratories such as NVIDIA, Intel, and FPGA suppliers engaged by CERN procurement. Key milestones included prototyping phases, integration tests during beam periods, and full deployment aligning with the LHC Run 3 start. The collaboration spans universities and institutes across Europe, Asia, and the Americas, including University of Frankfurt, University of Warsaw, Tata Institute of Fundamental Research, University of Tokyo, MIT, and University of California, Berkeley, organized under the ALICE collaboration governance structures and technical boards.

Category:ALICE experiment Category:CERN computing systems