A Software Data Transport Framework for Trigger Applications on Clusters

TL;DR

This paper presents a flexible, fault-tolerant software data transport framework designed for high data rate trigger applications on large clusters, demonstrating promising performance for CERN's ALICE experiment.

Contribution

It introduces a modular, runtime-configurable data transport framework with fault tolerance, tailored for high-throughput trigger systems in large-scale clusters.

Findings

Achieves high event rates suitable for ALICE requirements

Supports runtime reconfiguration of data transport components

Includes a basic fail-over mechanism for fault tolerance

Abstract

In the future ALICE heavy ion experiment at CERN's Large Hadron Collider input data rates of up to 25 GB/s have to be handled by the High Level Trigger (HLT) system, which has to scale them down to at most 1.25 GB/s before being written to permanent storage. The HLT system that is being designed to cope with these data rates consists of a large PC cluster, up to the order of a 1000 nodes, connected by a fast network. For the software that will run on these nodes a flexible data transport and distribution software framework has been developed. This framework consists of a set of separate components, that can be connected via a common interface, allowing to construct different configurations for the HLT, that are even changeable at runtime. To ensure a fault-tolerant operation of the HLT, the framework includes a basic fail-over mechanism that will be further expanded in the future, utilizing the runtime reconnection feature of the framework's component interface. First performance tests show very promising results for the software, indicating that it can achieve an event rate for the data transport sufficiently high to satisfy ALICE's requirements.