[Paper Review] Kubernetes as an Availability Manager for Microservice Applications

Leila Abdollahi Vayghan, Mohamed Aymen Saied+2 more|arXiv (Cornell University)|Jan 15, 2019

Software System Performance and Reliability12 references33 citations

TL;DR

The paper evaluates how Kubernetes manages availability for microservice apps, exploring architectures for public/private clouds, Kubernetes healing, and redundancy, and compares with the Availability Management Framework (AMF).

ABSTRACT

The move towards the microservice based architecture is well underway. In this architectural style, small and loosely coupled modules are developed, deployed, and scaled independently to compose cloud-native applications. However, for carrier-grade service providers to migrate to the microservices architectural style, availability remains a concern. Kubernetes is an open source platform that defines a set of building blocks which collectively provide mechanisms for deploying, maintaining, scaling, and healing containerized microservices. Thus, Kubernetes hides the complexity of microservice orchestration while managing their availability. In a preliminary work we evaluated Kubernetes, using its default configuration, from the availability perspective in a private cloud settings. In this paper, we investigate more architectures and conduct more experiments to evaluate the availability that Kubernetes delivers for its managed microservices. We present different architectures for public and private clouds. We evaluate the availability achievable through the healing capability of Kubernetes. We investigate the impact of adding redundancy on the availability of microservice based applications. We conduct experiments under the default configuration of Kubernetes as well as under its most responsive one. We also perform a comparative evaluation with the Availability Management Framework (AMF), which is a proven solution as a middleware service for managing high-availability. The results of our investigations show that in certain cases, the service outage for applications managed with Kubernetes is significantly high.

Motivation & Objective

• Motivate migration to microservices while ensuring availability in carrier-grade settings.
• Evaluate Kubernetes default and optimized configurations for availability in private and public cloud architectures.
• Assess the impact of adding redundancy on microservice application availability.
• Compare Kubernetes-based availability management with the Availability Management Framework (AMF).
• Analyze scenarios where Kubernetes healing effectiveness affects service continuity.

Proposed method

• Proposes architectures for deploying microservices in public and private clouds to study availability.
• Examines Kubernetes healing mechanisms as a means to maintain service availability.
• Performs experiments under Kubernetes default configuration and its most responsive configuration.
• Conducts a comparative evaluation with the Availability Management Framework (AMF) as a middleware solution for high-availability.
• Analyzes the impact of redundancy on availability in microservice deployments.

Experimental results

Research questions

• RQ1 How does Kubernetes healing influence the availability of managed microservices in different cloud architectures?
• RQ2 What is the impact of applying redundancy on the availability of microservice-based applications?
• RQ3 How does Kubernetes compare to the AMF in providing high-availability for microservices?
• RQ4 What differences arise between public and private cloud deployments regarding Kubernetes-enabled availability?
• RQ5 How do default versus most responsive Kubernetes configurations affect outage behavior?

Key findings

• In certain cases, service outage for Kubernetes-managed applications can be significantly high.
• Kubernetes healing capability contributes to availability, with varying effectiveness depending on configuration.
• Redundancy can alter availability outcomes, with observable improvements in some scenarios.
• Kubernetes’ availability performance differs from that of the AMF, indicating trade-offs between Kubernetes-native management and middleware-based approaches.
• Public versus private cloud architectures exhibit distinct availability characteristics under Kubernetes management.