Exploring polyglot software frameworks in ALICE with FairMQ and fer

TL;DR

This paper presents fer, a Go-based toolkit that interoperates with FairMQ, enabling polyglot distributed frameworks within the ALICE-O2 data processing environment to handle high data rates efficiently.

Contribution

Introduction of fer, a new Go-based toolkit that interoperates with FairMQ, facilitating polyglot programming in ALICE-O2's distributed data processing framework.

Findings

fer enables cross-language data exchange with FairMQ

Supports high data rates in ALICE-O2 framework

Demonstrates polyglot framework capabilities

Abstract

In order to meet the challenges of the Run 3 data rates and volumes, the ALICE collaboration is merging the online and offline infrastructures into a common framework: ALICE-O2. O2 is based on FairRoot and FairMQ, a message-based, multi-threaded and multi-process control framework. In FairMQ, processes (possibly on different machines) exchange data via message queues either through 0MQ or nanomsg. In turn, this enables developers to write their reconstruction or analysis process in whatever language they choose or deem appropriate for the task at hand, as long as that programming language can send and receive data through these message queues. This paper introduces fer, a Go-based toolkit that interoperates with the C++ toolkit FairMQ, to explore the realm of polyglot distributed frameworks.

Figures

Full text

1 Introduction

The ALICE collaboration is merging the online and offline infrastructures into a common framework: ALICE-$O^{2}$ ref-aliceo2 . This work is needed to meet the challenges of the Run 3 data rates and volumes. $O^{2}$ is a message-based, multi-threaded and multi-process control framework based on two libraries: FairRoot ref-fairroot and FairMQ ref-fairmq . Sophisticated processing pipelines can be modeled with FairMQ: router/dealer, request/reply, publish/subscribe, client/server, etc. The nodes of these pipelines are processes (possibly on different machines) exchanging data via message queues either through ZeroMQ ref-zeromq or nanomsg ref-nanomsg .

From the standpoint of the FairMQ toolkit, the programming language used to implement a given process in the pipeline is irrelevant, as long as data is correctly propagated through the message queues. This enables developers to write their reconstruction or analysis process in whatever language they choose or deem appropriate for the task at hand.

This paper presents fer, a Go-based ref-golang library compatible and interoperable with FairMQ. This paper starts with a brief introduction of the builtin features that make Go a solid choice when dealing with I/O and concurrency. The principal components of fer and how they interact with C++ FairMQ will then be described. Finally, Sec. 4 will report on the performance (CPU, VMem) of fer and conclude with the main figures of merit of fer, in the context of deployment in a distributed computing setup.

2 FairMQ concepts

FairMQ is a distributed processing toolkit, written in C++, with pluggable transports (ZeroMQ, nanomsg, shmem, InfiniBand). In FairMQ, the bulk of the processing is performed by devices: UNIX processes that are connected with other devices through message queues. Devices can be connected in various ways, through a wide variety of topologies: router/dealer, request/reply, publish/subscribe, client/server, etc. The medium used for the connection between devices can be chosen among the following options: tcp, udp, ipc, inproc, shared-memory. Figure 1 shows a typical FairMQ processing pipeline, fanning out data read off a source (sampler) to a pair of processors and then fanning in the data to a single output sink.

FairMQ applications are currently configured via a JSON configuration file, declaring the devices, the ports they are listening on or sending data to, the type and topology of the message queues, etc. An example of such a configuration file is shown in Fig. 2.