Orchestration Platform Research Articles

A Comparative Study of Resource Management Approaches in Kubernetes and Docker Swarm: Efficiency and Scalability

This article presents a comparative analysis of resource management (RM) approaches in two popular container orchestration platforms, Kubernetes and Docker Swarm. The key differences in RM, scheduling, and scaling are discussed, with a focus on the flexibility and granularity of Kubernetes compared to the simplicity and ease of use of Docker Swarm. The advantages and disadvantages of each tool are also analyzed to provide a more complete understanding of their applicability. -

A Framework for testing Federated Learning algorithms using an edge-like environment

Federated Learning (FL) is a machine learning paradigm in which many clients cooperatively train a single centralized model while keeping their data private and decentralized. FL is commonly used in edge computing, which involves placing computer workloads (both hardware and software) as close as possible to the edge, where data are created and where actions are occurring, enabling faster response times, greater data privacy, and reduced data transfer costs. However, due to the heterogeneous data distributions/contents of clients, it is non-trivial to accurately evaluate the contributions of local models in global centralized model aggregation. This is an example of a major challenge in FL, commonly known as data imbalance or class imbalance. In general, testing and evaluating FL algorithms can be a very difficult and complex task due to the distributed nature of the systems. In this work, a framework is proposed and implemented to evaluate FL algorithms in a more easy and scalable way. This framework is evaluated over a distributed edge-like environment managed by a container orchestration platform (i.e. Kubernetes).

An empirical study of developers’ challenges in implementing Workflows as Code: A case study on Apache Airflow

The Workflows as Code paradigm is becoming increasingly essential to streamline the design and management of complex processes within data-intensive software systems. These systems require robust capabilities to process, analyze, and extract insights from large datasets. Workflow orchestration platforms such as Apache Airflow are pivotal in meeting these needs, as they effectively support the implementation of the Workflows as Code paradigm. Nevertheless, despite its considerable advantages, developers still face challenges due to the specialized demands of workflow orchestration and the complexities of distributed execution environments. In this paper, we manually study 1000 sampled Stack Overflow posts derived from 9591 Airflow-related questions to understand developers’ challenges and root causes while implementing Workflows as Code. Our analysis results in a hierarchical taxonomy of Airflow-related challenges that contains 7 high-level categories and 14 sub-categories. We find that the most significant obstacles for developers arise when defining and executing their workflow. Our in-depth analysis identifies 10 root causes behind the challenges, including incorrect workflow configuration, complex environmental setup, and a lack of basic knowledge about Airflow and the external systems that it interacts with. Additionally, our analysis of references shared within the collected posts reveals that beyond the frequently cited Airflow documentation, documentation from external systems and third-party providers is also commonly referenced to address Airflow-related challenges.Editor’s note: Open Science material was validated by the Journal of Systems and Software Open Science Board.

Enhancing Edge Environment Scalability: Leveraging Kubernetes for Container Orchestration and Optimization

ABSTRACTKubernetes is an open‐source container orchestration platform, offers a comprehensive suite of features for managing containerized applications effectively. These features encompass horizontal scaling, per‐node‐pool cluster scaling and automated resource request adjustments. This research endeavors to harness these capabilities to address the limitations experienced by fog servers in edge environments, particularly those arising from restricted network connectivity and scalability challenges. In this research paper, the primary focus is on Kubernetes role of enhancing scalability, providing a robust framework for managing containerized applications. The proposed approach involves creating a predefined number of pods and containers within a Kubernetes cluster, specifically designed to efficiently handle incoming requests while optimizing CPU and memory usage. This method implements a microservice architecture for the web tier, with separate pods for the front end, back end and database, ensuring modular and scalable design. All pods communicate and integrate through REST APIs, facilitating seamless interaction and data exchange between the services. When handling web requests, the approach enables and controls both internal and external networks, ensuring secure and efficient communication. The analysis then examines the CPU and memory utilization of the pods, as well as node bandwidth, to provide a comprehensive evaluation of container scalability and performance within the Kubernetes cluster. These findings effectively demonstrate Kubernetes' capability in managing container scalability and optimizing resource utilization, highlighting its efficiency and robustness in a microservice environment.

Network Functions in Cloud: Kubernetes Deployment Challenges

The rapid evolution of cloud computing has paved the way for deploying network functions (NFs) in cloud environments, significantly enhancing the flexibility, scalability, and efficiency of modern network infrastructures. Kubernetes, an open-source container orchestration platform, has emerged as a leading tool for deploying and managing these cloud-based network functions. However, despite its widespread adoption, Kubernetes presents several deployment challenges specific to network functions, stemming from its design, scalability, and operational intricacies. This paper delves into the core challenges faced during the deployment of network functions on Kubernetes, focusing on issues related to network performance, security, service orchestration, and resource management. The abstract aims to provide an overview of the technical hurdles and propose potential strategies to overcome them, thus contributing to the optimization of Kubernetes-based NF deployments in cloud environments. By analyzing existing literature and case studies, the paper identifies key areas where improvements are needed and discusses the implications of these challenges for the future of cloud-based network functions. Ultimately, the paper seeks to guide network architects and cloud engineers in better understanding the complexities of Kubernetes deployments for network functions and in developing more effective strategies for successful implementation.

AI-Powered Anomaly Detection for Kubernetes Security: A Systematic Approach to Identifying Threats

This study delves into the intricacies of AI-based threat detection in Kubernetes security, with a specific focus on its role in identifying anomalous behavior. By harnessing the power of AI algorithms, vast amounts of telemetry data generated by Kubernetes clusters can be analyzed in real-time, enabling the identification of patterns and anomalies that may signify potential security threats or system malfunctions. The implementation of AI-based threat detection involves a systematic approach, encompassing data collection, model training, integration with Kubernetes orchestration platforms, alerting mechanisms, and continuous monitoring. AI-powered threat detection offers numerous advantages, including predictive threat detection, increased accuracy and scalability, shorter response times, and the ability to adapt to evolving threats. However, it also presents challenges, such as ensuring data quality, managing model complexity, mitigating false positives, addressing resource requirements, and maintaining security and privacy standards. The proposed AI-powered anomaly detection framework for Kubernetes security demonstrated significant improvements in threat identification and mitigation. Through real-time analysis of telemetry data and leveraging advanced AI algorithms, the system accurately identified over 92% of simulated security threats and anomalies across various Kubernetes clusters. Additionally, the integration of automated alerting mechanisms and response protocols reduced the average response time by 67%, enabling rapid containment of potential breaches.

Software-Bus-Toolchain (SBT): Introducing a Versatile Method for Quickly Implementing (I)IoT-Scenarios

The Internet of Things (IoT) has become ubiquitous. IoT devices are applied in a multitude of applications, e.g., in smart home scenarios, building automation, smart energy and smart cities, healthcare, and industrial environments. Fast and efficient implementation and roll-out of IoT devices is a critical factor for successs and acceptance of IoT devices. At the same time, the variety of hardware platforms that can be used for IoT applications, as well as the number of IoT orchestration platforms is increasing. Finding the right combination of tooling and hardware is not trivial, but essential for building applications that provide value. In this work, a Software-Bus-Toolchain (SBT) is introduced that encapsulates firmware design, data point definition, and communication protocol usage. Furthermore, an IoT control platform is provided to control and evaluate the IoT modules. Thus, using the SBT, solely the business logic has to be designed, while the hardware-design is automated to a high degree. Usage of the Zephyr framework allows the interchange of hardware modules, while interfaces provide easy adaption of data points and communication capabilities. The implementation of interfaces to the IoT-platform as well as to the communication layer provides a universal usage of logic and data elements. The SBT is evaluated in two application scenarios, where its flexible nature is shown.

Enhancing Resource Utilization Efficiency in Serverless Education: A Stateful Approach with Rofuse

Traditional container orchestration platforms often suffer from resource wastage in educational settings, and stateless serverless services face challenges in maintaining container state persistence during the teaching process. To address these issues, we propose a stateful serverless mechanism based on Containerd and Kubernetes, focusing on optimizing the startup process for container groups. We first implement a checkpoint/restore framework for container states, providing fundamental support for managing stateful containers. Building on this foundation, we propose the concept of “container groups” to address the challenges in educational practice scenarios characterized by a large number of similar containers on the same node. We then propose the Rofuse optimization mechanism, which employs delayed loading and block-level deduplication techniques. This enables containers within the same group to reuse locally cached file system data at the block level, thus reducing container restart latency. Experimental results demonstrate that our stateful serverless mechanism can run smoothly in typical educational practice scenarios, and Rofuse reduces the container restart time by approximately 50% compared to existing solutions. This research provides valuable exploration for serverless practices in the education domain, contributing new perspectives and methods to improve resource utilization efficiency and flexibility in teaching environments.

Taming latency at the edge: A user-aware service placement approach

Modern network and computing infrastructures are tasked with addressing the stringent demands of today’s applications. A pivotal concern is the minimization of latency experienced by end-users accessing services. While emerging network architectures provide a conducive setting for adept orchestration of microservices in terms of reliability, self-healing and resiliency, assimilating the awareness of the latency perceived by the user into placement decisions remains an unresolved problem. Current research addresses the problem of minimizing inter-service latency without any guarantee to the level of latency from the end-user to the cluster. In this research, we introduce an architectural approach for scheduling service workloads within a given cluster, prioritizing placement on the node that offers the lowest perceived latency to the end-user. To validate the proposed approach, we propose an implementation on Kubernetes (K8s), currently one of the most used workload orchestration platforms. Experimental results show that our approach effectively reduces the latency experienced by the end-user in a finite time without degrading the quality of service. We study the performance of the proposed approach analyzing different parameters with a particular focus on the size of the cluster and the number of replica pods involved to measure the latency. We provide insights on possible trade-offs between computational costs and convergence time.

OPTORER: A Dynamic Routing and Touring Service for Indoors and Outdoor Tours.

This paper introduces a new routing and touring service both for outdoor and indoor places of touristic and cultural interest designed to be used in the wider area of Attica, Greece. This service is the result of the work performed in OPTORER (OPTORER: OPtimal rouTing and explOration of touRistic and cultural arEas of interest within Attica given personalized adaptive preferences, promoted underlying purpose, and interactive experience), project, and it aspires to offer a range of innovative and thematic routes to several specified points of interest in the selected area of Attica, encouraging the combination of indoor and outdoor routes in a single tour. The aim is to optimize the user experience while promoting specific, user-centric features, with safety and social welfare being a priority for every designed tour, resulting in enhancing the touristic experience in the area. Using a common smartphone device, as well as common wearable devices (i.e., smartwatches), the OPTORER service will provide an end-to-end solution by developing the algorithms and end-user applications, together with an orchestration platform responsible for managing, operating, and executing the service that produces and presents to the end user results derived from solving dynamically complex optimization problems.

AI and agents

AbstractEarlier this year, OpenAI released their GPTs framework, allowing users to set up Large Language Model (LLM)‐based personas, orchestrate them into a workflow and even offering their AI apps within an app store. This is the latest, and maybe the easiest to set up, in a string of agent‐based LLM orchestration platforms in the past year, harkening a new age of agent‐based engineering. But, like most breakthroughs, this one is also rooted in many years of research, and the reason the world is paying attention to it now is that, thanks to Generative AI and Large Language Models, we finally have artificial agents that are useful enough to scale to more serious problems.

Impact of power consumption in containerized clouds: A comprehensive analysis of open-source power measurement tools

Recently, container-based solutions have become de facto compute units of modern cloud-native applications. However, the exponential growth in data traffic and the power consumption of these technologies to handle high data traffic alarm the strong need for energy evaluation approaches in containerized clouds. Furthermore, the proliferation of highly distributed edge clouds raises additional concerns regarding the power consumption of future cloud architectures. This article presents a detailed overview of methods and techniques for monitoring power consumption within popular cloud platforms. The study offers an in-depth evaluation of these approaches, demonstrating variations in measured power consumption based on the chosen technique. A well-known container orchestration platform named Kubernetes (K8s) has been applied in our extensive measurements. This work argues that energy-efficient container clouds will play a vital role in building a more sustainable and eco-friendly digital infrastructure by optimizing power consumption and reducing carbon footprint, paving the way for a greener future. The paper also discusses open challenges and future research directions on energy sustainability, leading to the conclusion, offering lessons learned and prospects on potential solutions to foster sustainable practices within the container ecosystem.

Design of Mixed Reality Visualization System for Operational Situation Using Cloud-based Geospatial Information

The importance of geospatial information is increasingly highlighted in the defense domain. Accurate and up-to-date geospatial data is essential for situational awareness, target analysis, and mission planning in millitary operations. The use of high-resolution geospatial data in military operations requires large storage and fast image processing capabilities. Efficient image processing is required for tasks such as extracting useful information from satellite images and creating 3D terrain for mission planning, In this paper, we designed a cloud-based operational situation mixed reality visualization system that utilizes large-scale geospatial information distributed processed on a cloud server based on the container orchestration platform Kubernetes. We implemented a prototype and confirmed the suitability of the design.

Applying a System of Systems Perspective to Hyundai‐Kia's Virtual Tire Development

ABSTRACTSystems engineering has become important in almost every complex product manufacturing industry, especially automotive. Emerging trends like vehicle electrification and autonomous driving now pose a system of systems (SoS) engineering challenge to automotive OEMs. This paper presents a proof‐of‐concept (PoC) that applies a top‐down SoS perspective to Hyundai‐Kia Motor Corporation's (HKMC) virtual product development process to develop a performance‐critical component of the vehicle, the tire. The PoC demonstrates using the Arcadia MBSE method to develop a consistent, layered, vehicle architecture model starting from the SoS operational context down to the lowest level of system decomposition in the physical architecture thereby capturing top‐down knowledge traceability Using the concept of functional chains, several vehicle performance views are captured that serve as the basis for architecture verification orchestration across engineering domains using a cross‐domain orchestration platform thereby validating key vehicle/tire performance metrics that influence the tire design parameters. Preliminary results of the study show that applying a method‐based modeling approach could provide several benefits to HKMC's current product development approach such as reduced time to model, SoS knowledge capture and reusability, parameter/requirement traceability, early performance verification, and effective systems engineering collaboration between the OEM, tire design supplier, and tire manufacturers.

5G-VIOS: Towards next generation intelligent inter-domain network service orchestration and resource optimisation

This paper introduces an intelligent network service orchestration platform, referred to as 5G-VIOS for 5G networks and beyond in accordance with the Zero-touch Network and Service Management (ZSM) paradigm. The proposed solution is responsible for an automated network slicing and life-cycle management of network applications and services, across multiple administrative and technological domains. An Artificial Intelligent (AI) model utilising Machine Learning (ML) techniques is exploited to intelligently and efficiently profile the network services and predict the efficient configuration of resources needed to meet the performance targets and Service Level Agreements (SLAs) of these network services across multiple domains. We test and validate the performance of the prediction models for both resource configuration and utilisation in various settings for different resources and data rates. We also showcase how resource utilisation predictions of a virtualised network service can significantly assist in its life cycle management by proactively preventing unnecessary actions such as its migration.

Comparison of application container orchestration platforms

This article presents a comparative analysis of three well-known container orchestration platforms: Docker Swarm, Kubernetes and Apache Mesos, focusing on the deployment of a test application and measuring parameters such as deployment time, memory, CPU and disk utilization, application response time and the time to restore a replica of the application using an auto-recovery mechanism. The aim of the research is to verify the performance and efficiency of the analyzed platforms, facilitating informed decisions while choosing an orchestrator for containerized applications. Two research hypotheses have been stated. The first one assumes that the time required to launch an application using the Docker Swarm tool is the shortest among the analyzed platforms. The second hypothesis is that Kubernetes provides the most efficient results in terms of load scheduling and application scaling. The analysis performed on the Jenkins application showed the superiority of the Docker Swarm platform over the other studied tools in terms of performance.

Gossip-Based Monitoring Protocol for 6G Networks

The service function (SF) area has gained increasing attention in the last years due its ability to combine the advantages of cloud computing with network softwarization. By decoupling SFs from the physical equipment where they are executed, it is possible to make network services scalable and flexible. These advantages become even more evident in the forthcoming 6G networks, where the overall environment is expected to become more dynamic and cloud-based, with SFs deployed as cloud-native functions. However, in order to efficiently manage and compose services using these SFs, it is necessary to monitor the available resources of the nodes where they can be deployed, in addition to exchange information relevant to the operational status of active SFs. To this aim, we propose a lightweight monitoring architecture by using agents in charge of monitoring the status of SFs running in co-located clusters. These monitoring agents exchange their information by means of a gossip protocol, which allows increasing the reliability of the process. In this way, it is possible to keep service decisions as local as possible, limiting the interactions with centralized decision and orchestration platforms, and thus increasing network scalability and responsiveness. Performance evaluation shows the effectiveness of the proposed solution, and demonstrates that the network overhead of the distributed monitoring process is definitely affordable.

Diktyo: Network-Aware Scheduling in Container-Based Clouds

Containers have revolutionized application deployment and life-cycle management in current cloud platforms. Applications have evolved from single monoliths to complex graphs of loosely-coupled microservices. However, the efficient allocation of microservice-based applications is challenging due to their complex inter-dependencies. Further, recent applications are becoming even more delay-sensitive, demanding lower latency between dependent microservices. Scheduling policies in popular container orchestration platforms mainly aim to increase the resource efficiency of the infrastructure, insufficient for latency-sensitive applications. Application domains such as the Internet of Things and multi-tier web services would benefit from network-aware policies that consider network latency and bandwidth in the scheduling process. Previous works have studied network-aware scheduling via theoretical formulations or heuristic-based methods evaluated via simulations or small testbeds, making their full applicability in popular platforms difficult. This paper proposes a novel network-aware framework for the popular Kubernetes (K8s) platform named Diktyo that determines the placement of dependent microservices in long-running applications focused on reducing the application’s end-to-end latency and guaranteeing bandwidth reservations. Simulations show that Diktyo can significantly reduce the network latency for various applications across different infrastructure topologies compared to default K8s scheduling plugins. Also, experiments in a K8s cluster with microservice benchmark applications show that Diktyo can increase database throughput by 22% and reduce application response time by 45%.

Research of the efficiency of computing services management platforms in the organization of fog computing

The work is devoted to studying the effectiveness of computing service management platforms in the organization of Fog Computing. As part of the work, the effectiveness of container orchestration platforms with the Fog computing organization is being studied. During the research, it is necessary to complete the following tasks: 1) select literature, scientific publications and Internet articles necessary for the research; 2) inspect container orchestration platforms; 3) determine the key requirements and criteria for conducting the study; 4) design and implement an automatic testing utility; 5) conduct a study of the effec-tiveness of container orchestration platforms with the organization of fog computing; 6) ana-lyze the results obtained and draw related conclusions. Deployment of Docker containers is organized. Docker Swarm is used to create a clus-ter. The problems of measuring the following parameters are solved: deployment time of one container, deployment time of a group of containers, response time of the horizontal zoom task, transmission delay time. The analysis of the obtained test results is carried out.

Toward IoT Node Authentication Mechanism in Next Generation Networks

Although the next generation networks (5G-NGNs) provide a flexible infrastructure to support latency-sensitive and bandwidth-hungry mission-critical Internet of Things (IoT) applications, however, the 5G-IoT integration in NGNs has increased the threat surface. Unfortunately, IoT devices are resource constrained, and the traditional Intrusion Detection Systems (IDS) approaches based on cryptography are not effective on 5G-IoT ecosystems. In this paper, we propose an effective 5GIoT node authentication approach that leverages unique RF (Radio Frequency) fingerprinting data to train the Deep learning model to detect legitimate and non-legitimate IoT nodes. Our approach is based on Mahalanobis Distance theory and Chisquare distribution theories. The proposed approach achieves a higher detection accuracy (99.35%) as well as lower training time compared to other existing approaches which is a key benefit of our approach in NGNs. The experiments are conducted using ETSI-Open Source NFV Management and Orchestration (OSM-MANO) platform on AWS (Amazon Web Services) cloud platform to verify how the proposed approach would fit in reallife scenarios. The method can be used as a standalone security system or as a part of multi-factor authentication.