Network-Aware Reliability Modeling and Optimization for Microservice Placement

TL;DR

This paper introduces a network-aware reliability model and placement algorithms for microservices, significantly improving service reliability and bandwidth efficiency in IoT and mobile networks.

Contribution

It presents a novel reliability model considering network load and routing, along with efficient placement algorithms and shared backup mechanisms for microservices.

Findings

SRP algorithm reduces service failures by up to 29%.

SRP-S algorithm cuts bandwidth use by up to 62%.

Shared backup paths improve reliability and efficiency.

Abstract

Optimizing microservice placement to enhance the reliability of services is crucial for improving the service level of microservice architecture-based mobile networks and Internet of Things (IoT) networks. Despite extensive research on service reliability, the impact of network load and routing on service reliability remains understudied, leading to suboptimal models and unsatisfactory performance. To address this issue, we propose a novel network-aware service reliability model that effectively captures the correlation between network state changes and reliability. Based on this model, we formulate the microservice placement problem as an integer nonlinear programming problem, aiming to maximize service reliability. Subsequently, a service reliability-aware placement (SRP) algorithm is proposed to solve the problem efficiently. To reduce bandwidth consumption, we further discuss the microservice placement problem with the shared backup path mechanism and propose a placement algorithm based on the SRP algorithm using shared path reliability calculation, known as the SRP-S algorithm. Extensive simulations demonstrate that the SRP algorithm reduces service failures by up to 29% compared to the benchmark algorithms. By introducing the shared backup path mechanism, the SRP-S algorithm reduces bandwidth consumption by up to 62% compared to the SRP algorithm with the fully protected path mechanism. It also reduces service failures by up to 21% compared to the SRP algorithm with the shared backup mechanism.