Towards Sustainable Computing: Exploring Energy Consumption Efficiency of Alternative Configurations and Workloads in an Open Source Messaging System Part of the research leading to the results presented in this paper has received funding from the European Union’s funded Project HUMAINE under grant

Maria Voreakou ∗ , George Kousiouris, Mara Nikolaidou · 2025

Read Towards Sustainable Computing: Exploring Energy Consumption Efficiency of Alternative Configurations and Workloads in an Open Source Messaging System Part of the research leading to the results presented in this paper has received funding from the European Union’s funded Project HUMAINE under grant by Maria Voreakou ∗ , George Kousiouris, Mara Nikolaidou (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-10693.txt.

Introduction

Maria Voreakou, George Kousiouris, and Mara Nikolaidou meter the electricity appetite of a messaging backbone in a 2025 paper. Message brokers quietly carry much of modern computing's traffic, shuttling events among microservices and devices. Their energy draw, across configurations, had rarely been benchmarked head to head.

The experiments run RabbitMQ, a leading open-source broker, under varied workloads, message rates, and consumer counts, and under alternative system architectures. Architectural choice alone moved power draw by up to thirty-one percent, a margin worth real money and carbon at data-center scale. The team publishes the full dataset openly, aiming to feed energy-aware cost models and greener default configurations. The broader point lands beyond one product: plumbing choices, often made by habit, are climate choices at scale.

Catalog record

CREATORMaria Voreakou ∗ , George Kousiouris, Mara Nikolaidou
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-10693

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Towards Sustainable Computing: Exploring Energy Consumption Efficiency of Alternative Configurations and Workloads in an Open Source Messaging System Part of the research leading to the results presented in this paper has received funding from the European Union’s funded Project HUMAINE under grant agreement No. 101120218. Maria Voreakou ∗ , George Kousiouris, Mara Nikolaidou Abstract Energy consumption in current large scale computing infrastructures is becoming a critical issue, especially with the growing demand for centralized systems such as cloud environments. With the advancement of microservice architectures and the Internet of Things, messaging systems have become an integral and mainstream part of modern computing infrastructures, carrying out significant workload in a majority of applications. In this paper, we describe an experimental process to explore energy-based benchmarking for RabbitMQ, one of the main open source messaging frameworks. The involved system is described, as well as required components, and setup scenarios, involving different workloads and configurations among the tests as well as messaging system use cases. Alternative architectures are investigated and compared from an energy consumption point of view, for different message rates and consumer numbers. Differences in architectural selection have been quantified and can lead to up to 31% reduction in power consumption. The resulting dataset is made publicly available and can thus prove helpful for architectures’ comparison, energy-based cost modeling, and beyond. I Introduction Sustainability and Green computing [ 29 ] are important topics of the current era, with multiple and complex applications across large data centers, consuming more and more energy as an integral part of our digital life. Data centers’ energy consumption has started to become an issue in recent years [ 20 ] [ 30 ] . Therefore, careful consideration needs to be applied on the design, configuration, and operation of systems and software under various loads, in order to reduce energy consumption, by closely monitoring and analyzing different strategies, optimizations, and understanding the behavior of such systems. Messaging systems have emerged as one of the main backbones of modern application and system architectures. Messaging systems act as a main backbone interconnecting different systems, in a System of Systems approach, in order to enforce greater separation, enhance scalability and fault tolerance, as well as enable abstraction and functional separation between systems. They are extensively used either as data distribution systems (e.g. in IoT platforms [ 34 ] ), generic load balancing mechanisms (e.g. in the case of the queue-based load leveling pattern [ 31 ] ), as core cloud services (e.g. Amazon SNS/SQS) or as the main mechanisms in social networks [ 32 ] . Therefore, they are highly relevant in a distributed, large-scale, and cloud native context, carrying out significant workload in day-to-day activities. They also ensure message delivery in faulty environments with out of the box functionalities for message tracking, delivery guarantees and cross-system notifications. The scope of this study is to extract the energy footprint of one of the most commonly used open source messaging systems, RabbitMQ, under varying incoming and outgoing loads, represented through the parameters of input messages per second, number of routing keys as well as number of consumer queues. Furthermore, we take into consideration different strategies in its configuration, such as the selection of the main type of exchange for the same usage scenario. RabbitMQ consists of 3 different exchanges: Direct , Topic , and Fanout Exchange, which define the way the messages are distributed to the consumers through the queues. Direct creates a simple Exchange with a strict filtering pattern on the message, so the consumers subscribed to this exchange receive all the messages sent to this queue that exactly match the desired routing key. Similarly, Topic creates an exchange with a partial or full match filtering pattern, and the consumers subscribed to the specific topic can consume the messages from the bound queues, if the partial or full match is successful. Fanout on the other hand is an exchange which broadcasts the message to all the queues that are bound to this exchange, so all the consumers listening to any queue of this exchange will consume all the messages. The goals of this experiment are as follows: • Create a testbed architecture that includes the main components (load generator, RabbitMQ setup, energy monitoring, etc.) and define scenarios to explore, by setting parameters such as message rate, Queues, Routing keys, etc. • Map the different experimentation parameters on the energy needed for the operation of the system. The analysis of such data may result in understanding how these parameters affect the total energy consumed. Having that information may result in future specific energy-based billing models for messaging systems offered as a service, i.e. in cloud-based environments. • Quantify the effect of choosing between two different strategies in order to distribute messages based on routing keys, as shown in Figure 1 . While the typical strategy within RabbitMQ is to use one topic-based exchange with full key matching or a direct exchange, an alternative design may be considered in which each key is represented by a separate fanout exchange. In this case, all interested consumers are bound to each of the desired exchanges. With this, we achieve eliminating routing key matching, and thus we could potentially see different behavior in terms of energy consumption. As a trade-off, we need to have a more complex registration process and potentially larger memory needs due to the multiple exchanges’ design. Fig. 1: RabbitMQ Direct vs Fanout Exchange Experiment Strategy Based on the overall results and the relevant conclusions, a systems engineer should be able to define the best possible setups so that the application created is both more sustainable and better performing. The paper is structured as follows. In Section 2, the relevant works are mentioned, in terms of relevant tools and analysis. In Section 3, we present the Project Testbed with the selected tools, including hardware and software setup, the test application and the experiment scenarios we are going to run. In Section 4, we present an initial analysis of the results including the main observed patterns and trends. Finally, in Section 5, we conclude the analysis and define future steps. II Related Work Authors of [ 24 ] built a decentralized MQTT for a real-time remote healthcare monitoring system using local edge devices, and measured the energy consumption using the turbostat tool, to determine the overall system energy footprint. No special information regarding the RabbitMQ energy consumption is mentioned. However, they concluded that RabbitMQ has proved resilient in managing data during network disruptions, ensuring uninterrupted patient monitoring. Authors of [ 27 ] made an analysis in the current carbon estimation methodologies and proposed a usage and allocation based carbon model. Although this model aligns only with serverless computing, they concluded that it is valuable to design models based on carbon-aware pricing which is influenced directly by the power consumption. To enforce such approaches on messaging systems, the authors proposed that relevant benchmark data such as the ones presented in this work are needed. Authors of [ 21 ] evaluated existing energy models ( PowerAPI [ 22 ] and Scaphandre [ 17 ] ) in servers using different performance settings such as hyperthreading [ 8 ] and turboboost [ 9 ] to measure the

Read the complete text