Container Orchestration Systems Research Articles

Efficient Resource Allocation in Kubernetes Using Machine Learning

Kubernetes is a distinguished open-source container orchestration system in cloud computing and containerized applications. Google developed it, and the Cloud Native Computing Foundation now maintains it. Kubernetes offers a robust framework for automating application deployment scaling and management, revolutionizing how organizations use their containerized workloads and providing huge flexibility and feasibility. The current paper explores the application of machine learning algorithms to optimize resource allocation in Kubernetes environments. As the complexity of cloud-native applications increases because of various engagements, it is vital to maintain performance and cost- effectiveness. This study also evaluates various machine learning models and techniques and their relevancy in areas such as anomaly detection and enhancing overall cluster utilization. Our findings include machine learning-driven methodologies that will significantly improve performance utilizing historical data. Kubernetes's decentralized nature requires a scalable structure for task scheduling to accommodate dynamic workloads conveniently. The AIMD algorithm, a celebrated method for congestion avoidance in network management, inspires our approach. Computing clusters can be challenging to deploy and manage due to their complexity. Monitoring systems collect large amounts of data, which is daunting to understand manually. Machine learning provides a viable solution to detect anomalies in a Kubernetes cluster. KAD (Kubernetes et al.) is one such algorithm that can solve the Cluster anomalies problem. Enormous Cloud native applications market is projected to reach 17.0 billion USD by 2028, which was USD 5.9 billion in 2023. On par with those numbers is the global Machine Learning (ML) market size, which was valued at $19.20 billion in 2022 and is expected to grow from $26.03 billion in 2023 to $225.91 billion by 2030 (As per Fortune Business Insights). At the conjecture, both markets will take innovation to a new level, offering more adaptive solutions for contemporary cloud infrastructures.

SLA-Adaptive Threshold Adjustment for a Kubernetes Horizontal Pod Autoscaler

Kubernetes is an open-source container orchestration system that provides a built-in module for dynamic resource provisioning named the Horizontal Pod Autoscaler (HPA). The HPA identifies the number of resources to be provisioned by calculating the ratio between the current and target utilisation metrics. The target utilisation metric, or threshold, directly impacts how many and how quickly resources will be provisioned. However, the determination of the threshold that would allow satisfying performance-based Service Level Objectives (SLOs) is a long, error-prone, manual process because it is based on the static threshold principle and requires manual configuration. This might result in underprovisioning or overprovisioning, leading to the inadequate allocation of computing resources or SLO violations. Numerous autoscaling solutions have been introduced as alternatives to the HPA to simplify the process. However, the HPA is still the most widely used solution due to its ease of setup, operation, and seamless integration with other Kubernetes functionalities. The present study proposes a method that utilises exploratory data analysis techniques along with moving average smoothing to identify the target utilisation threshold for the HPA. The objective is to ensure that the system functions without exceeding the maximum number of events that result in a violation of the response time defined in the SLO. A prototype was created to adjust the threshold values dynamically, utilising the proposed method. This prototype enables the evaluation and comparison of the proposed method with the HPA, which has the highest threshold set that meets the performance-based SLOs. The results of the experiments proved that the suggested method adjusts the thresholds to the desired service level with a 1–2% accuracy rate and only 4–10% resource overprovisioning, depending on the type of workload.

Fast state transfer for updates and live migration of industrial controller runtimes in container orchestration systems

Many industrial processes (e.g., steel, fertilizer, paper production) utilize programmable logic controllers or distributed control systems, which cyclically execute control applications. Dynamic updates of such applications at runtime are desirable for optimizing the production while avoiding production stops. However, such updates are challenging since they require fast switching to an updated instance and executing an internal state transfer within the control cycle times. Existing dynamic update approaches for industrial applications can for example not update controller runtimes or operating systems, since they do not utilize modern container virtualization technologies. Furthermore, existing state transfer methods for container orchestration systems do not support fast cyclic applications. We propose a novel state transfer method for industrial control applications deployed as containerized Kubernetes microservices on server hosts or resource-rich industrial personal computers. The method enables dynamic updates and live migration of applications to others hosts without interrupting the underlying production facilities, which avoids costly downtimes. We have implemented our method as a Kubernetes extension and found that it can transfer large application states (i.e. 100.000 variables) via OPC UA in less than 42 ms across different hosts, which can fit into the available cycle slack time, therefore enabling a wide range of bumpless updates at runtime.

Approach to Evaluate Scheduling Strategies in Container Orchestration Systems

This article offers an in-depth examination of scheduling strategies in container orchestration systems, emphasizing the importance of selecting the most suitable strategy for a system's specific needs. The article explores the evolution of container systems and challenges around cloud reliability and fault tolerance. A structured experimental methodology was introduced to compare different strategies systematically, utilizing various testing methods like packing until first rejection, packing until the cluster is full, and introducing different malfunctions such as liveness and network partitioning. Comparative results between the “binpack” and “spread” strategies highlighted that while the former excels in higher scheduling requests acceptance rate, the latter is superior in offering higher availability. The findings show the need for aligning the scheduling strategy with system objectives, be it resource optimization, fault tolerance or other desired characteristics

Scalable decarbonisation using automated operation optimisation

One of the biggest challenges in facing the climate crisis is the decarbonization of the large and diverse building stock. A reduction of carbon dioxide emissions can be achieved by technical measures and engaging the building occupants to adapt their behaviour. Among the technical measures, implementing predictive control as an upgrade of the existing heating, ventilation, air conditioning and cooling system is especially promising as it allows reductions at potentially low running cost. However, the effort for adapting, implementing and deploying these methods to fit specific buildings and scenarios is high and requires special domain knowledge, hindering the wide-spread application. In this paper, we present a highly automated and data-driven implementation process utilizing an open-source container orchestration system, and the results from real-life case studies in existing buildings in which predictive control was retrofitted. Additionally, occupant information systems were installed in the buildings for increasing transparency about the building performance and the effect of the occupants’ behaviour. The shown method is useful for reducing the time required and manual effort for implementing new control strategies, and thus reducing carbon dioxide emissions while simultaneously increasing thermal comfort and air quality.

Autonomous workload distribution for container-based micro services environments

Microservice architecture represents a cloud application design approach that transfers the intricacies from the conventional monolithic applications to the infrastructure. It involves breaking down the application into small, containerized microservices, each responsible for a specific functional requirement. These microservices can be independently deployed, scaled, and tested through automated orchestration systems. Our paper introduces an autonomous system for distributing workloads among containerized microservices within the cloud like a swarm, designed specifically for microservices operating within OpenVZ containers. This system has the potential to enhance performance compared to existing centralized container orchestration systems.

A Kubernetes dataset for misuse detection

Container security involves a broad spectrum of concerns, including the security of the operating system, auditing the supply chain and the application security of the running containers. This wide attack surface will also include the security of the container orchestration system and its components once a container orchestration system is introduced to manage the fleet of containers in an environment. In order to advance the research in this field, prior work should be comparable and reproducible. However, we identified a research gap for this aspect; publicly available datasets for container security is sparse and reproducibility of the research output so far is arduous. In this study, we share a dataset consisting of network flows, collected from a Kubernetes cluster. Furthermore, we performed a preliminary analysis on the data as a sanity check to evaluate its quality. By sharing this dataset publicly, we hope to help further studies and establish benchmarks in the field of container networking security.

Rescheduling Strategy for Container Orchestration System to Improve Application Availability

Virtualization technologies such as containers have become increasingly popular and widely used today. Containers offer several advantages, such as flexibility, portability, and scalability. Container-supporting technologies such as container orchestration provide more advantages to container users to maintain the containers. One of the crucial features of container orchestration is the scheduler responsible for assigning containers to the proper server. However, container misplacement often occurs during the scheduling process, which may reduce the application's Quality of Service (QoS), such as availability. Multiple studies on scheduling strategies have been conducted with various goals, such as increasing application availability. To improve the application availability, we can reschedule the containers in the cluster. This study proposes a rescheduling strategy on the Kubernetes container orchestration system to improve the application available in the container. We use the probability approach as a solution for rescheduling actions in dynamic environments against containers that have an unhealthy state. Our experimental results show that our rescheduling strategy improves application availability compared with the Kubernetes default schedule. Rescheduling improves application availability according to experiments. After rescheduling, Kubernetes had a better application success rate. The chart demonstrates that the average success rate of applications for the threshold of 0.9 surpasses the other thresholds on average. In contrast, the availability of applications using the threshold 0f 0.7 is lower than the others. High thresholds improve application availability.

EOS deployment on Ceph RBD/CephFS with K8s

The present activity focused on the integration of different storage systems (EOS [1] and Ceph [2]) with the aim to combine the high level scalability and stability of EOS services with the reliability and redundancy features provided by Ceph. The work has been carried out as part of the collaboration between the national center of INFN (Italian Institute for Nuclear Physics) dedicated to Research and Development on Information and Communication Technologies and the Conseil Européen pour la Recherche Nucléaire (CERN), with the goal of evaluating and testing different technologies for next-generation storage challenges. This work leverages the well-known open-source container orchestration system, Kubernetes [3], for managing file system services. The results obtained by measuring the performances of the different combined technologies, comparing for instance block device and file system as backend options provided by a Ceph cluster deployed on physical machines, are shown and discussed in the manuscript.

Kubernetes Scheduling: Taxonomy, Ongoing Issues and Challenges

Continuous integration enables the development of microservices-based applications using container virtualization technology. Container orchestration systems such as Kubernetes, which has become the de facto standard, simplify the deployment of container-based applications. However, developing efficient and well-defined orchestration systems is a challenge. This article focuses specifically on the scheduler, a key orchestrator task that assigns physical resources to containers. Scheduling approaches are designed based on different Quality of Service (QoS) parameters to provide limited response time, efficient energy consumption, better resource utilization, and other things. This article aims to establish insight knowledge into Kubernetes scheduling, find the main gaps, and thus guide future research in the area. Therefore, we conduct a study of empirical research on Kubernetes scheduling techniques and present a new taxonomy for Kubernetes scheduling. The challenges, future direction, and research opportunities are also discussed.

Machine Learning-based Orchestration of Containers: A Taxonomy and Future Directions

Containerization is a lightweight application virtualization technology, providing high environmental consistency, operating system distribution portability, and resource isolation. Existing mainstream cloud service providers have prevalently adopted container technologies in their distributed system infrastructures for automated application management. To handle the automation of deployment, maintenance, autoscaling, and networking of containerized applications, container orchestration is proposed as an essential research problem. However, the highly dynamic and diverse feature of cloud workloads and environments considerably raises the complexity of orchestration mechanisms. Machine learning algorithms are accordingly employed by container orchestration systems for behavior modeling and prediction of multi-dimensional performance metrics. Such insights could further improve the quality of resource provisioning decisions in response to the changing workloads under complex environments. In this article, we present a comprehensive literature review of existing machine learning-based container orchestration approaches. Detailed taxonomies are proposed to classify the current researches by their common features. Moreover, the evolution of machine learning-based container orchestration technologies from the year 2016 to 2021 has been designed based on objectives and metrics. A comparative analysis of the reviewed techniques is conducted according to the proposed taxonomies, with emphasis on their key characteristics. Finally, various open research challenges and potential future directions are highlighted.

Benchmarking geospatial database on Kubernetes cluster

Kubernetes is an open-source container orchestration system for automating container application operations and has been considered to deploy various kinds of container workloads. Traditional geo-databases face frequent scalability issues while dealing with dense and complex spatial data. Despite plenty of research work in the comparison of relational and NoSQL databases in handling geospatial data, there is a shortage of existing knowledge about the performance of geo-database in a clustered environment like Kubernetes. This paper presents benchmarking of PostgreSQL/PostGIS geospatial databases operating on a clustered environment against non-clustered environments. The benchmarking process considers the average execution times of geospatial structured query language (SQL) queries on multiple hardware configurations to compare the environments based on handling computationally expensive queries involving SQL operations and PostGIS functions. The geospatial queries operate on data imported from OpenStreetMap into PostgreSQL/PostGIS. The clustered environment powered by Kubernetes demonstrated promising improvements in the average execution times of computationally expensive geospatial SQL queries on all considered hardware configurations compared to their average execution times in non-clustered environments.

Enabling scalable and fault-tolerant multi-agent systems by utilizing cloud-native computing

Multi-agent systems (MAS) represent a distributed computing paradigm well suited to tackle today’s challenges in the field of the Internet of Things (IoT). Both share many similarities such as the interconnection of distributed devices and their cooperation. The combination of MAS and IoT would allow the transfer of the experience gained in MAS research to the broader range of IoT applications. The key enabler for utilizing MAS in the IoT is the ability to build large-scale and fault-tolerant MASs since IoT concepts comprise possibly thousands or even millions of devices. However, well known multi-agent platforms (MAP), e. g., Java Agent DE-velopment Framework (JADE), are not able to deal with these challenges. To this aim, we present a cloud-native Multi-Agent Platform (cloneMAP) as a modern MAP based on cloud-computing techniques to enable scalability and fault-tolerance. A microservice architecture is used to implement it in a distributed way utilizing the open-source container orchestration system Kubernetes. Thereby, bottlenecks and single-points of failure are conceptually avoided. A comparison with JADE via relevant performance metrics indicates the massively improved scalability. Furthermore, the implementation of a large-scale use case verifies cloneMAP’s suitability for IoT applications. This leads to the conclusion that cloneMAP extends the range of possible MAS applications and enables the integration with IoT concepts.

Optimized container scheduling for data-intensive serverless edge computing

Operating data-intensive applications on edge systems is challenging, due to the extreme workload and device heterogeneity, as well as the geographic dispersion of compute and storage infrastructure. Serverless computing has emerged as a compelling model to manage the complexity of such systems, by decoupling the underlying infrastructure and scaling mechanisms from applications. Although serverless platforms have reached a high level of maturity, we have found several limiting factors that inhibit their use in an edge setting. This paper presents a container scheduling system that enables such platforms to make efficient use of edge infrastructures. Our scheduler makes heuristic trade-offs between data and computation movement, and considers workload-specific compute requirements such as GPU acceleration. Furthermore, we present a method to automatically fine-tune the weights of scheduling constraints to optimize high-level operational objectives such as minimizing task execution time, uplink usage, or cloud execution cost. We implement a prototype that targets the container orchestration system Kubernetes, and deploy it on an edge testbed we have built. We evaluate our system with trace-driven simulations in different infrastructure scenarios, using traces generated from running representative workloads on our testbed. Our results show that (a) our scheduler significantly improves the quality of task placement compared to the state-of-the-art scheduler of Kubernetes, and (b) our method for fine-tuning scheduling parameters helps significantly in meeting operational goals.

Container description ontology for CaaS

Besides its classical three service models (IaaS, PaaS, and SaaS), container as a service (CaaS) has gained significant acceptance. It offers without the difficulty of high-performance challenges of traditional hypervisors deployable applications. As the adoption of containers is increasingly wide spreading, the use of tools to manage them across the infrastructure becomes a vital necessity. In this paper, we propose a conceptualisation of a domain ontology for the container description called CDO. CDO presents, in a detailed and equal manner, the functional and non-functional capabilities of containers, Dockers and container orchestration systems. In addition, we provide a framework that aims at simplifying the container management not only for the users but also for the cloud providers. In fact, this framework serves to populate CDO, help the users to deploy their application on a container orchestration system, and enhance interoperability between the cloud providers by providing migration service for deploying applications among different host platforms. Finally, the CDO effectiveness is demonstrated relying on a real case study on the deployment of a micro-service application over a containerised environment under a set of functional and non-functional requirements.

Hard drives monitoring automation approach for Kubernetes container orchestration system

Today, a laborious and non-trivial task is to automate monitoring of hard drives in a cluster infrastructure using the Kubernetes container management system. The paper discusses existing approaches to monitoring hard drives in the Kubernetes container orchestration system and provides a comparative analysis of them. Based on the presented analysis, a conclusion is drawn on the need to improve and automate approaches. The paper proposes an approach to automating the collection of metrics from hard drives by implementing the Kubernetes “operator” for a tool with which you can effectively obtain information about the state of hard drives in the system. As results, the temporal characteristics of collecting information about disks using existing approaches and the proposed approach are given. Numerical results and graphs showing the gain of the proposed approach are presented

Container description ontology for CaaS

Besides its classical three service models (IaaS, PaaS, and SaaS), container as a service (CaaS) has gained significant acceptance. It offers without the difficulty of high-performance challenges of traditional hypervisors deployable applications. As the adoption of containers is increasingly wide spreading, the use of tools to manage them across the infrastructure becomes a vital necessity. In this paper, we propose a conceptualisation of a domain ontology for the container description called CDO. CDO presents, in a detailed and equal manner, the functional and non-functional capabilities of containers, Dockers and container orchestration systems. In addition, we provide a framework that aims at simplifying the container management not only for the users but also for the cloud providers. In fact, this framework serves to populate CDO, help the users to deploy their application on a container orchestration system, and enhance interoperability between the cloud providers by providing migration service for deploying applications among different host platforms. Finally, the CDO effectiveness is demonstrated relying on a real case study on the deployment of a micro-service application over a containerised environment under a set of functional and non-functional requirements.

Analysis Preservation and Systematic Reinterpretation within the ATLAS experiment

The LHC data analysis software used in order to derive and publish experimental results is an important asset that is necessary to preserve in order to fully exploit the scientific potential of a given measurement. An important use-case is the re-usability of the analysis procedure in the context of new scientific studies such as the reinterpretation of searches for new physics in terms of signal models that not studied in the original publication (RECAST). We present the usage of the graph-based workflow description language yadage to drive the reinterpretation of preserved HEP analyses. The analysis software is preserved using Docker containers, while the workflow structure is preserved using plain JSON documents. This allows the re-execution of complex analysis workflows on modern distributed container orchestration systems and enables a systematic reinterpretation service based on such preserved analysis.

Dynamic Multi-level Auto-scaling Rules for Containerized Applications

Container-based cloud applications require sophisticated auto-scaling methods in order to operate under different workload conditions. The choice of an auto-scaling method may significantly affect important service quality parameters, such as response time and resource utilization. Current container orchestration systems such as Kubernetes and cloud providers such as Amazon EC2 employ auto-scaling rules with static thresholds and rely mainly on infrastructure-related monitoring data, such as CPU and memory utilization. This paper presents a new dynamic multi-level (DM) auto-scaling method with dynamically changing thresholds, which uses not only infrastructure, but also application-level monitoring data. The new method is compared with seven existing auto-scaling methods in different synthetic and real-world workload scenarios. Based on experimental results, all eight auto-scaling methods are compared according to the response time and the number of instantiated containers. The results show that the proposed DM method has better overall performance under varied amount of workloads than the other auto-scaling methods. Due to satisfactory results, the proposed DM method is implemented in the SWITCH software engineering system for time-critical cloud applications.

Time Complexity of an Distributed Algorithm for Load Balancing of Microservice-oriented Applications in the Cloud

Microservice architecture is a relatively new cloud application design pattern. Each microservice has a single responsibility in terms of functional requirement, and that can be managed independently from other microservices. This is done using automated cloud orchestration systems. In this paper we analyze the time complexity of an simple swarm-like decentralized load balancing algorithm for microservices running inside OpenVZ virtualization containers. We show that it can offer performance improvements with respect to the existing centralized container orchestration systems.