arXiv subjects

Yehia Elkhatib

Publications and source records attributed to Yehia Elkhatib.

At least 19 recordsLinked to original sources

arXiv2026

How Accurately Can the Energy Use of Spark Applications Be Estimated Based on Resource Utilisation?

Distributed batch data processing applications are widely executed on cloud-based resources where restricted user access to node-level hardware energy counters hinders transparent sustainability accounting. Energy and carbon attribution methodologies therefore depend on power models and available resource utilisation traces, yet the accuracy of these estimates has to be validated while direct counters are available. In this work, we use Apache Spark running on Kubernetes as a case-study dataflow runtime and cluster resource manager to compare model-based energy estimates to Intel RAPL package and DRAM energy on an AWS bare-metal cloud and an on-premises cluster, comparing different CPU usage signals and memory coefficients. We show that external monitoring improves signed package-energy error relative to Spark task traces, reducing underestimation from -29.58% to -24.41% on AWS and from -24.00% to -16.22% on-premises.

cs.DC↗

arXiv2026

Ichnos+: Estimating the Carbon Footprint of Scientific Workflows Using Fitted Power Models

As data-intensive scientific workflows scale to facilitate the automation of analysis of increasing amounts of data, their resource-intensive and long-running execution incurs significant energy consumption and carbon emissions. Given the already significant and rising emissions from the ICT sector, it is crucial to quantify and understand the carbon footprint of scientific workflows. However, existing tooling is commonly not usable in shared, virtualized environments or resorts to power models that are based on only one or two generic data points. To address this gap, this paper presents Ichnos+, a novel system to quantify the environmental footprint of Nextflow scientific workflows. Ichnos+ enables post-hoc footprint estimation based on existing workflow traces, node-specific power models for the computational resources utilized, and carbon intensity data aligned with the execution time. We evaluate Ichnos+ against hardware-level energy measurements obtained using Intel RAPL, and the nf-core co2footprint plugin, which implements the Green Algorithms methodology. We find that Ichnos+ is capable of estimating workflow energy consumption with an estimation error of 10.8% across three compute clusters, significantly outperforming the nf-core plugin. We further show that Ichnos+ extends beyond operational carbon to estimate embodied emissions as well as water and land use. Finally, we demonstrate how Ichnos+ can be extended for another workflow system, Apache Airflow, maintaining a similarly high degree of estimation accuracy.

cs.DC↗

arXiv2026

Augur: Pre-Execution Energy Prediction for Workflow Tasks in Heterogeneous Clusters

Scientific workflows are widely used to process large quantities of data, leading to significant energy consumption and carbon emissions. To reduce this environmental impact, energy and carbon-aware scheduling approaches could be employed. However, such methods require runtime and energy predictions, which are typically only available for workflows that have been executed previously. Meanwhile, scientists may execute new or modified workflows, use workflows with different input data, or run them on alternative infrastructure. To address this critical gap, we propose Augur, a novel method to predict the energy consumption of scientific workflow tasks prior to execution. By efficiently profiling both the available cluster infrastructure and the workflow at hand, Augur is capable of predicting the overall energy consumption of the workflow with a median prediction error of $16.3\pm15.3\%$ compared to Ichnos, an energy estimation method that uses fitted power models, and $18.2\pm14.7\%$ compared to Intel RAPL, as observed in our experimental evaluation on public and private cloud infrastructure. Relying on only minimal historical execution data, Augur outperforms two state-of-the-art methods in predicting both task runtime and total workflow energy, providing a robust foundation for energy-efficient and carbon-aware scientific data analysis.

cs.DC↗

arXiv2026

A Systematic Evaluation of the Potential of Carbon-Aware Execution for Scientific Workflows

Scientific workflows are critical to scientific data analysis and often involve computationally intensive processing of large datasets on compute clusters. As such, their execution tends to be long-running and resource-intensive, resulting in significant energy consumption and carbon emissions. While carbon-aware computing methods have received considerable attention in general cloud contexts, their application to scientific data analysis workflows remains a critical research gap. Our study addresses this oversight by showing how the delay tolerance, interruptibility, and scalability of scientific workflows can be leveraged for a significantly more sustainable execution model. In this study, we first quantify the problem of carbon emissions associated with running scientific workflows, and then demonstrate the transformative potential for carbon-aware workflow execution. We estimate the carbon footprint of seven real-world Nextflow workflows executed on diverse dedicated cluster and public cloud resources using high-resolution average and marginal grid carbon intensity data from open and commercial data providers. Furthermore, we conduct a systematic evaluation of the impact of carbon-aware temporal shifting, and the dynamic pausing and resuming of the workflow. Moreover, we investigate the impact of resource scaling at both workflow and workflow task levels. Finally, we report substantial potential reductions in overall carbon emissions, with temporal shifting capable of decreasing emissions by over 80%, and resource scaling by 67%.

cs.DC↗

arXiv2025

xMem: A CPU-Based Approach for Accurate Estimation of GPU Memory in Deep Learning Training Workloads

The global scarcity of GPUs necessitates more sophisticated strategies for Deep Learning jobs in shared cluster environments. Accurate estimation of how much GPU memory a job will require is fundamental to enabling advanced scheduling and GPU sharing, which helps prevent out-of-memory (OOM) errors and resource underutilization. However, existing estimation methods have limitations. Approaches relying on static analysis or historical data with machine learning often fail to accurately capture runtime dynamics. Furthermore, direct GPU analysis consumes scarce resources, and some techniques require intrusive code modifications. Thus, the key challenge lies in precisely estimating dynamic memory requirements, including memory allocator nuances, without consuming GPU resources and non-intrusive code changes. To address this challenge, we propose xMem, a novel framework that leverages CPU-only dynamic analysis to accurately estimate peak GPU memory requirements a priori. We conducted a thorough evaluation of xMem against state-of-the-art solutions using workloads from 25 different models, including architectures like Convolutional Neural Networks and Transformers. The analysis of 5209 runs, which includes ANOVA and Monte Carlo results, highlights xMem's benefits: it decreases the median relative error by 91% and significantly reduces the probability of estimation failure as safe OOM thresholds by 75%, meaning that the estimated value can often be used directly without causing OOM. Ultimately, these improvements lead to a 368% increase in memory conservation potential over current solutions.

cs.PF↗

arXiv2025

Ichnos: A Carbon Footprint Estimator for Scientific Workflows

Scientific workflows facilitate the automation of data analysis, and are used to process increasing amounts of data. Therefore, they tend to be resource-intensive and long-running, leading to significant energy consumption and carbon emissions. With ever-increasing emissions from the ICT sector, it is crucial to quantify and understand the carbon footprint of scientific workflows. However, existing tooling requires significant effort from users - such as setting up power monitoring before executing workloads, or translating monitored metrics into the carbon footprints post-execution. In this paper, we introduce a system to estimate the carbon footprint of Nextflow scientific workflows that enables post-hoc estimation based on existing workflow traces, power models for computational resources utilised, and carbon intensity data aligned with the execution time. We discuss our automated power modelling approach, and compare it with commonly used estimation methodologies. Furthermore, we exemplify several potential use cases and evaluate our energy consumption estimation approach, finding its estimation error to be between 3.9-10.3%, outperforming both baseline methodologies.

cs.DC↗

arXiv2025

Energy-Aware Workflow Execution: An Overview of Techniques for Saving Energy and Emissions in Scientific Compute Clusters

Scientific research in many fields routinely requires the analysis of large datasets, and scientists often employ workflow systems to leverage clusters of computers for their data analysis. However, due to their size and scale, these workflow applications can have a considerable environmental footprint in terms of compute resource use, energy consumption, and carbon emissions. Mitigating this is critical in light of climate change and the urgent need to reduce carbon emissions. In this chapter, we exemplify the problem by estimating the carbon footprint of three real-world scientific workflows from different scientific domains. We then describe techniques for reducing the energy consumption and, thereby, carbon footprint of individual workflow tasks and entire workflow applications, such as using energy-efficient heterogeneous architectures, generating optimised code, scaling processor voltages and frequencies, consolidating workloads on shared cluster nodes, and scheduling workloads for optimised energy efficiency.

cs.DC↗

arXiv2025

Accurate GPU Memory Prediction for Deep Learning Jobs through Dynamic Analysis

The benefits of Deep Learning (DL) impose significant pressure on GPU resources, particularly within GPU cluster, where Out-Of-Memory (OOM) errors present a primary impediment to model training and efficient resource utilization. Conventional OOM estimation techniques, relying either on static graph analysis or direct GPU memory profiling, suffer from inherent limitations: static analysis often fails to capture model dynamics, whereas GPU-based profiling intensifies contention for scarce GPU resources. To overcome these constraints, VeritasEst emerges. It is an innovative, entirely CPU-based analysis tool capable of accurately predicting the peak GPU memory required for DL training tasks without accessing the target GPU. This "offline" prediction capability is core advantage of VeritasEst, allowing accurate memory footprint information to be obtained before task scheduling, thereby effectively preventing OOM and optimizing GPU allocation. Its performance was validated through thousands of experimental runs across convolutional neural network (CNN) models: Compared to baseline GPU memory estimators, VeritasEst significantly reduces the relative error by 84% and lowers the estimation failure probability by 73%. VeritasEst represents a key step towards efficient and predictable DL training in resource-constrained environments.

cs.DC↗

arXiv2025

Exploring the Potential of Carbon-Aware Execution for Scientific Workflows

Scientific workflows are widely used to automate scientific data analysis and often involve processing large quantities of data on compute clusters. As such, their execution tends to be long-running and resource intensive, leading to significant energy consumption and carbon emissions. Meanwhile, a wealth of carbon-aware computing methods have been proposed, yet little work has focused specifically on scientific workflows, even though they present a substantial opportunity for carbon-aware computing because they are inherently delay tolerant, efficiently interruptible, and highly scalable. In this study, we demonstrate the potential for carbon-aware workflow execution. For this, we estimate the carbon footprint of two real-world Nextflow workflows executed on cluster infrastructure. We use a linear power model for energy consumption estimates and real-world average and marginal CI data for two regions. We evaluate the impact of carbon-aware temporal shifting, pausing and resuming, and resource scaling. Our findings highlight significant potential for reducing emissions of workflows and workflow tasks.

cs.DC↗

arXiv2021

Disaggregated Memory at the Edge

This paper describes how to augment techniques such as Distributed Shared Memory with recent trends on disaggregated Non Volatile Memory in the data centre so that the combination can be used in an edge environment with potentially volatile and mobile resources. This article identifies the main advantages and challenges, and offers an architectural evolution to incorporate recent research trends into production-ready disaggregated edges. We also present two prototypes showing the feasibility of this proposal.

cs.DC↗

arXiv2020

SLO-ML: A Language for Service Level Objective Modelling in Multi-cloud Applications

Cloud modelling languages (CMLs) are designed to assist customers in tackling the diversity of services in the cloud market. While many CMLs have been proposed in the literature, they lack practical support for automating the selection of services based on the specific service level objectives of a customer's application. We put forward SLO-ML, a novel and generative CML to capture service level requirements and, subsequently, to select the services to honour customer requirements and generate the deployment code appropriate to these services. We present the architectural design of SLO-ML and the associated broker that realises the deployment operations. We rigorously evaluate SLO-ML using a mixed methods approach. First, we exploit an experimental case study with a group of researchers and developers using a real-world cloud application. We also assess overheads through an exhaustive set of empirical scalability tests. Through expressing the levels of gained productivity and experienced usability, we highlight SLO-ML's profound potential in enabling user-centric cloud brokers. We also discuss limitations as application requirements grow.

cs.DC↗

arXiv2019

Optimizing Deep Learning Inference on Embedded Systems Through Adaptive Model Selection

Deep neural networks ( DNNs ) are becoming a key enabling technology for many application domains. However, on-device inference on battery-powered, resource-constrained embedding systems is often infeasible due to prohibitively long inferencing time and resource requirements of many DNNs. Offloading computation into the cloud is often unacceptable due to privacy concerns, high latency, or the lack of connectivity. While compression algorithms often succeed in reducing inferencing times, they come at the cost of reduced accuracy. This paper presents a new, alternative approach to enable efficient execution of DNNs on embedded devices. Our approach dynamically determines which DNN to use for a given input, by considering the desired accuracy and inference time. It employs machine learning to develop a low-cost predictive model to quickly select a pre-trained DNN to use for a given input and the optimization constraint. We achieve this by first off-line training a predictive model, and then using the learned model to select a DNN model to use for new, unseen inputs. We apply our approach to two representative DNN domains: image classification and machine translation. We evaluate our approach on a Jetson TX2 embedded deep learning platform and consider a range of influential DNN models including convolutional and recurrent neural networks. For image classification, we achieve a 1.8x reduction in inference time with a 7.52% improvement in accuracy, over the most-capable single DNN model. For machine translation, we achieve a 1.34x reduction in inference time over the most-capable single model, with little impact on the quality of translation.

cs.LG↗

arXiv2019

IoTNetSim: A Modelling and Simulation Platform for End-to-End IoT Services and Networking

Internet-of-Things (IoT) systems are becoming increasingly complex, heterogeneous and pervasive, integrating a variety of physical devices and virtual services that are spread across architecture layers (cloud, fog, edge) using different connection types. As such, research and design of such systems have proven to be challenging. Despite the influx in IoT research and the significant benefits of simulation-based approaches in supporting research, there is a general lack of appropriate modelling and simulation platforms to create a detailed representation of end-to-end IoT services, i.e. from the underlying IoT nodes to the application layer in the cloud along with the underlying networking infrastructure. To aid researchers and practitioners in overcoming these challenges, we propose IoTNetSim, a novel self-contained extendable platform for modelling and simulation of end-to-end IoT services. The platform supports modelling heterogeneous IoT nodes (sensors, actuators, gateways, etc.) with their fine-grained details (mobility, energy profile, etc.), as well as different models of application logic and network connectivity. The proposed work is distinct from the current literature, being an all-in-one tool for end-to-end IoT services with a multi-layered architecture that allows modelling IoT systems with different structures. We experimentally validate and evaluate our IoTNetSim implementation using two very large-scale real-world cases from the natural environment and disaster monitoring IoT domains.

cs.DC↗

arXiv2019

Transferable Knowledge for Low-cost Decision Making in Cloud Environments

Users of cloud computing are increasingly overwhelmed with the wide range of providers and services offered by each provider. As such, many users select cloud services based on description alone. An emerging alternative is to use a decision support system (DSS), which typically relies on gaining insights from observational data in order to assist a customer in making decisions regarding optimal deployment or redeployment of cloud applications. The primary activity of such systems is the generation of a prediction model (e.g. using machine learning), which requires a significantly large amount of training data. However, considering the varying architectures of applications, cloud providers, and cloud offerings, this activity is not sustainable as it incurs additional time and cost to collect training data and subsequently train the models. We overcome this through developing a Transfer Learning (TL) approach where the knowledge (in the form of the prediction model and associated data set) gained from running an application on a particular cloud infrastructure is transferred in order to substantially reduce the overhead of building new models for the performance of new applications and/or cloud infrastructures. In this paper, we present our approach and evaluate it through extensive experimentation involving three real world applications over two major public cloud providers, namely Amazon and Google. Our evaluation shows that our novel two-mode TL scheme increases overall efficiency with a factor of 60\% reduction in the time and cost of generating a new prediction model. We test this under a number of cross-application and cross-cloud scenarios.

cs.DC↗

arXiv2019

Cloud Futurology

The Cloud has become integral to most Internet-based applications and user gadgets. This article provides a brief history of the Cloud and presents a researcher's view of the prospects for innovating at the infrastructure, middleware, and application and delivery levels of the already crowded Cloud computing stack.

cs.DC↗

arXiv2018

On Improving Service Chains Survivability through Efficient Backup Provisioning

With the growing adoption of Software Defined Networking (SDN) and Network Function Virtualization (NFV), large-scale NFV infrastructure deployments are gaining momentum. Such infrastructures are home to thousands of network Service Function Chains (SFCs), each composed of a chain of virtual network functions (VNFs) that are processing incoming traffic flows. Unfortunately, in such environments, the failure of a single node may break down several VNFs and thereby breaking many service chains at the same time. In this paper, we address this particular problem and investigate possible solutions to ensure the survivability of the affected service chains by provisioning backup VNFs that can take over in case of failure. Specifically, we propose a survivability management framework to efficiently manage SFCs and the backup VNFs. We formulate the SFC survivability problem as an integer linear program that determines the minimum number of required backups to protect all the SFCs in the system and identifies their optimal placement in the infrastructure. We also propose two heuristic algorithms to cope with the large-scale instances of the problem. Through extensive simulations of different deployment scenarios, we show that these algorithms provide near-optimal solutions with minimal computation time.

cs.NI↗

arXiv2018

Navigating Diverse Data Science Learning: Critical Reflections Towards Future Practice

Data Science is currently a popular field of science attracting expertise from very diverse backgrounds. Current learning practices need to acknowledge this and adapt to it. This paper summarises some experiences relating to such learning approaches from teaching a postgraduate Data Science module, and draws some learned lessons that are of relevance to others teaching Data Science.

cs.GL↗

arXiv2018

Research Challenges in Nextgen Service Orchestration

Fog/edge computing, function as a service, and programmable infrastructures, like software-defined networking or network function virtualisation, are becoming ubiquitously used in modern Information Technology infrastructures. These technologies change the characteristics and capabilities of the underlying computational substrate where services run (e.g. higher volatility, scarcer computational power, or programmability). As a consequence, the nature of the services that can be run on them changes too (smaller codebases, more fragmented state, etc.). These changes bring new requirements for service orchestrators, which need to evolve so as to support new scenarios where a close interaction between service and infrastructure becomes essential to deliver a seamless user experience. Here, we present the challenges brought forward by this new breed of technologies and where current orchestration techniques stand with regards to the new challenges. We also present a set of promising technologies that can help tame this brave new world.

cs.DC↗