Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PBScaler: A Bottleneck-aware Autoscaling Framework for Microservice-based Applications

Shuaiyu Xie, Jian Wang, Bing Li, Zekun Zhang, Duantengchuan Li, Patrick C. K. H

TL;DR

PBScaler addresses the challenge of autoscaling in microservice architectures by focusing on bottleneck localization to prevent performance degradation. It introduces TopoRank, a topological-potential based random walk that combines anomaly potential with Personalized PageRank to identify PBs from a microservice dependency graph, followed by offline GA-based scaling optimization guided by an SLO-violation predictor. The approach reduces unnecessary scaling, limits oscillations, and improves end-to-end performance while cutting resource consumption, as demonstrated on real-world and emulated workloads. The combination of metric-based observability, offline optimization, and bottleneck-centric decisions provides a practical, scalable solution for cloud-native microservices.

Abstract

Autoscaling is critical for ensuring optimal performance and resource utilization in cloud applications with dynamic workloads. However, traditional autoscaling technologies are typically no longer applicable in microservice-based applications due to the diverse workload patterns and complex interactions between microservices. Specifically, the propagation of performance anomalies through interactions leads to a high number of abnormal microservices, making it difficult to identify the root performance bottlenecks (PBs) and formulate appropriate scaling strategies. In addition, to balance resource consumption and performance, the existing mainstream approaches based on online optimization algorithms require multiple iterations, leading to oscillation and elevating the likelihood of performance degradation. To tackle these issues, we propose PBScaler, a bottleneck-aware autoscaling framework designed to prevent performance degradation in a microservice-based application. The key insight of PBScaler is to locate the PBs. Thus, we propose TopoRank, a novel random walk algorithm based on the topological potential to reduce unnecessary scaling. By integrating TopoRank with an offline performance-aware optimization algorithm, PBScaler optimizes replica management without disrupting the online application. Comprehensive experiments demonstrate that PBScaler outperforms existing state-of-the-art approaches in mitigating performance issues while conserving resources efficiently.

PBScaler: A Bottleneck-aware Autoscaling Framework for Microservice-based Applications

TL;DR

PBScaler addresses the challenge of autoscaling in microservice architectures by focusing on bottleneck localization to prevent performance degradation. It introduces TopoRank, a topological-potential based random walk that combines anomaly potential with Personalized PageRank to identify PBs from a microservice dependency graph, followed by offline GA-based scaling optimization guided by an SLO-violation predictor. The approach reduces unnecessary scaling, limits oscillations, and improves end-to-end performance while cutting resource consumption, as demonstrated on real-world and emulated workloads. The combination of metric-based observability, offline optimization, and bottleneck-centric decisions provides a practical, scalable solution for cloud-native microservices.

Abstract

Autoscaling is critical for ensuring optimal performance and resource utilization in cloud applications with dynamic workloads. However, traditional autoscaling technologies are typically no longer applicable in microservice-based applications due to the diverse workload patterns and complex interactions between microservices. Specifically, the propagation of performance anomalies through interactions leads to a high number of abnormal microservices, making it difficult to identify the root performance bottlenecks (PBs) and formulate appropriate scaling strategies. In addition, to balance resource consumption and performance, the existing mainstream approaches based on online optimization algorithms require multiple iterations, leading to oscillation and elevating the likelihood of performance degradation. To tackle these issues, we propose PBScaler, a bottleneck-aware autoscaling framework designed to prevent performance degradation in a microservice-based application. The key insight of PBScaler is to locate the PBs. Thus, we propose TopoRank, a novel random walk algorithm based on the topological potential to reduce unnecessary scaling. By integrating TopoRank with an offline performance-aware optimization algorithm, PBScaler optimizes replica management without disrupting the online application. Comprehensive experiments demonstrate that PBScaler outperforms existing state-of-the-art approaches in mitigating performance issues while conserving resources efficiently.
Paper Structure (25 sections, 14 equations, 13 figures, 4 tables, 3 algorithms)

This paper contains 25 sections, 14 equations, 13 figures, 4 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Bottleneck Analysis
  4. Autoscaling for Microservices
  5. System Design
  6. Metric Collector
  7. Performance Bottleneck Analysis
  8. SLO Violation Detection
  9. Redundancy Checking
  10. Bottleneck Localization
  11. Scaling Decision
  12. Constrained Optimization Model
  13. SLO Violation Predictor
  14. Autoscaling Optimization
  15. EVALUATIONS
  16. ...and 10 more sections

Figures (13)

  • Figure 1: Part of the invocation relationship in Online Boutique.
  • Figure 2: Latency distribution in four scenarios.
  • Figure 3: The replica fluctuation and latency fluctuation of MicroScaler under burst workloads. Excessive online attempts (a) cause oscillations and performance degradation (b).
  • Figure 4: Framework of PBScaler.
  • Figure 5: Visualization of query results for metric istio_requests_total.
  • ...and 8 more figures