Horizontal Pod Autoscaling in Kubernetes for Elastic Container Orchestration

Nguyen, Thanh-Tung; Yeom, Yu Jin; Kim, Taejoon; Park, Dae Heon; Kim, Se-Han

doi:10.3390/s20164621

Cited by 102 publications

(49 citation statements)

References 19 publications

(24 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In a Kubernetes-based environment, it is referenced by increasing or decreasing the number of replicas of the same POD. When an application requires more computational resources, instead of having to adjust the specifications of the existing pods, users can simply create another identical pod to share the load [Nguyen et al 2020]. In an environment purely using Docker containers (without any orchestration framework), it would be analogous to refer only to the addition or reduction of one or more containers of the same service.…”

Section: Types Of Microservice Autoscalingmentioning

confidence: 99%

“…Preferably, a complete method should be able to identify the best time to autoscale resources either horizontally or vertically. By fully exploiting elasticity, an application can more quickly react to small workload variations, through finegrained vertical scaling, as well as to sudden workload peaks, through horizontal scaling [Nguyen et al 2020].…”

Section: Types Of Microservice Autoscalingmentioning

confidence: 99%

See 1 more Smart Citation

State of the Art on Microservices Autoscaling: An Overview

Nunes¹,

Bianchi²,

Iwasaki³

et al. 2021

Anais Do XLVIII Seminário Integrado De Software E Hardware (SEMISH 2021)

View full text Add to dashboard Cite

The adoption of microservices architecture has taken on great pro-portions due to its benefits and popularization of containers driven tools, such as Kubernetes and Docker. Besides, the development of microservice-based applications is a complex task, specially because they can be composed of multiple heterogeneous parts. In particular, one of the main challenges is how to conduct the microservices autoscaling (i.e., adding or removing resources on demand), while still avoiding resource waste, such as CPU and memory. This paper presents the state of the art of approaches to solve the problem of micro services autoscaling, the main characteristics to be considered as well as the important future directions that need to be still investigated.

show abstract

Section: Types Of Microservice Autoscalingmentioning

confidence: 99%

Section: Types Of Microservice Autoscalingmentioning

confidence: 99%

State of the Art on Microservices Autoscaling: An Overview

Nunes¹,

Bianchi²,

Iwasaki³

et al. 2021

Anais Do XLVIII Seminário Integrado De Software E Hardware (SEMISH 2021)

View full text Add to dashboard Cite

show abstract

“…The goal is to expand the capacity of the application before the peak load by adjusting the number of pods in advance according to the prediction result. In [ 9 ], the authors investigate the horizontal pod autoscaler (HPA), which is one of the most important features in Kubernetes. They conduct various experiments to deeply analyze HPA based on several metrics, such as Kubernetes resource metrics (e.g., CPU and memory usage) and Prometheus custom metrics (e.g., the average arrival rate of HTTP requests) [ 10 ].…”

Section: Related Workmentioning

confidence: 99%

Balanced Leader Distribution Algorithm in Kubernetes Clusters

Nguyen

Kim

2021

Sensors

Self Cite

View full text Add to dashboard Cite

Container-based virtualization is becoming a de facto way to build and deploy applications because of its simplicity and convenience. Kubernetes is a well-known open-source project that provides an orchestration platform for containerized applications. An application in Kubernetes can contain multiple replicas to achieve high scalability and availability. Stateless applications have no requirement for persistent storage; however, stateful applications require persistent storage for each replica. Therefore, stateful applications usually require a strong consistency of data among replicas. To achieve this, the application often relies on a leader, which is responsible for maintaining consistency and coordinating tasks among replicas. This leads to a problem that the leader often has heavy loads due to its inherent design. In a Kubernetes cluster, having the leaders of multiple applications concentrated in a specific node may become a bottleneck within the system. In this paper, we propose a leader election algorithm that overcomes the bottleneck problem by evenly distributing the leaders throughout nodes in the cluster. We also conduct experiments to prove the correctness and effectiveness of our leader election algorithm compared with a default algorithm in Kubernetes.

show abstract

“…Kubernetes [8][9][10] offers Horizontal Pod Autoscaler (HPA) [11][12][13], a built-in horizontal scaling controller, which automatically scales the ReplicaSet controller, deployment controller, or pod quantity based on statistical CPU utilization (or other custom metrics). is section presents the Kubernetes' horizontal scaling technique, including the acquisition of HPA metrics, how it works, and its limitations.…”

Section: Related Workmentioning

confidence: 99%

A Fine-Grained Horizontal Scaling Method for Container-Based Cloud

Jiang

2021

Scientific Programming

View full text Add to dashboard Cite

The container scaling mechanism, or elastic scaling, means the cluster can be dynamically adjusted based on the workload. As a typical container orchestration tool in cloud computing, Horizontal Pod Autoscaler (HPA) automatically adjusts the number of pods in a replication controller, deployment, replication set, or stateful set based on observed CPU utilization. There are several concerns with the current HPA technology. The first concern is that it can easily lead to untimely scaling and insufficient scaling for burst traffic. The second is that the antijitter mechanism of HPA may cause an inadequate number of onetime scale-outs and, thus, the inability to satisfy subsequent service requests. The third concern is that the fixed data sampling time means that the time interval for data reporting is the same for average and high loads, leading to untimely and insufficient scaling at high load times. In this study, we propose a Double Threshold Horizontal Pod Autoscaler (DHPA) algorithm, which fine-grained divides the scale of events into three categories: scale-out, no scale, and scale-in. And then, on the scaling strength, we also employ two thresholds that are further subdivided into no scaling (antijitter), regular scaling, and fast scaling for each of the three cases. The DHPA algorithm determines the scaling strategy using the average of the growth rates of CPU utilization, and thus, different scheduling policies are adopted. We compare the DHPA with the HPA algorithm under different loads, including low, medium, and high. The experiments show that the DHPA algorithm has better antijitter and antiload characteristics in container increase and reduction while ensuring service and cluster security.

show abstract

Horizontal Pod Autoscaling in Kubernetes for Elastic Container Orchestration

Cited by 102 publications

References 19 publications

State of the Art on Microservices Autoscaling: An Overview

State of the Art on Microservices Autoscaling: An Overview

Balanced Leader Distribution Algorithm in Kubernetes Clusters

A Fine-Grained Horizontal Scaling Method for Container-Based Cloud

Contact Info

Product

Resources

About