1,720,982 research outputs found
Optimizing the integration of agent-based cloud archestrators and higher-level workloads
No full textFog native architecture : intent-based workflows to take cloud native toward the edge
Full text linkThe cloud native approach is rapidly transforming how applications are developed and operated, turning monolithic applications into microservice applications, allowing teams to release faster, increase reliability, and expedite operations by taking full advantage of cloud resources and their elasticity. At the same time, "fog computing" is emerging, bringing the cloud toward the edge, near the end user, in order to increase privacy, improve resource efficiency, and reduce latency. Combining these two trends, however, proves difficult because of four fundamental disconnects between the cloud native paradigm and fog computing. This article identifies these disconnects and proposes a fog native architecture along with a set of design patterns to take full advantage of the fog. Central to this approach is turning microservice applications into microservice workflows, constructed dynamically by the system using an intent-based approach taking into account a number of factors such as user requirements, request location, and available infrastructure and microservices. The architecture introduces a novel softwarized fog mesh facilitating both inter-microservice connectivity, external communication, and end-user aggregation. Our evaluation analyzes the impact of distributing microservice-based applications over a fog ecosystem, illustrating the impact of CPU and network latency and application metrics on perceived quality of service of fog native workflows compared to the cloud. The results show the fog can offer superior application performance given the right conditions
Cloud-native-bench : an extensible benchmarking framework to streamline cloud performance tests
Full text linkThe shift to the cloud among organizations has surged enormously over the last decade. As a result, an overwhelming amount of new cloud-based technologies emerged, making it increasingly more challenging to compare the different technologies and to identify the ideal technology that aligns best with a specific use case. The performance and resource usage of a system or software technology can be assessed by running benchmarks. Performing benchmarks has proven to be a time-consuming and error-prone task, especially when executing multiple consecutive tests on the same system. This sparked curiosity in exploring the feasibility of automating this manual benchmarking process. This paper proposes Cloud-NativeBench (CNB), a novel open source benchmarking framework implemented as a Kubernetes operator that fully automates the benchmarking cycle. The entire process, including benchmark deployment, the consecutive execution of benchmarks in a queue, results collection, and statistical data analysis, is fully automated. The framework is designed to be extensible without the need to adapt the operator itself, enabling users to develop fine-tuned custom benchmarks according to their specific use cases. A detailed evaluation shows the ease-of-use of Cloud-Native-Bench and how it streamlines the process of running benchmarks in cloud-native environments. Experiments show the importance of running benchmarks on cloud technologies. For example, employing a different web server technology can increase the mean throughput by 252%
Edge anomaly detection framework for AIOps in Cloud and IoT
Full text linkArtificial Intelligence for IT Operations (AIOps) addresses the rising complexity of cloud computing and Internet of Things by assisting DevOps engineers to monitor and maintain applications. Machine Learning is an essential part of AIOps, enabling it to perform Anomaly Detection and Root Cause Analysis. These techniques are often executed in centralized components, however, which requires transferring vast amounts of data to a central location. This increase in network traffic causes strain on the network and results in higher latency. This paper leverages edge computing to address this issue by deploying ML models closer to the monitored services, reducing the network overhead. This paper investigates two architectural approaches: a sidecar architecture and a federated architecture, and highlights their advantages and shortcomings in different scenarios. Taking this into account, it proposes a framework that orchestrates the deployment and management of distributed edge ML models. Additionally, the paper introduces a Python library to assist data scientists during the development of AIOps techniques and concludes with a thorough evaluation of the resulting framework towards resource consumption and scalability. The results indicate up to 98.3% reduction in network usage depending on the configuration used while maintaining a minimal increase in resource usage at the edge
Beyond generic lifecycles: Reusable modeling of custom-fit management workflows for cloud applications
No full textUnikernels vs containers: An in-depth benchmarking study in the context of microservice applications
No full text
Solid Web Monetization
Full text linkThis research was partially funded by Grant for the Web, a fund to boost open, fair, and inclusive standards and innovation in Web Monetization. -Ruben Taelman is a postdoctoral fellow of the Research Foundation -Flanders (FWO) (1274521N)
- …
