Operations 9 min read

Efficient Online Performance Testing Using a Mirror Cluster Self‑Service Platform

This article describes how Ctrip’s senior testing manager designed a self‑service online performance testing solution based on a Mirror cluster, detailing its architecture, implementation steps, safety measures, result aggregation, current limitations, and overall impact on testing efficiency and reliability.

Ctrip Technology
Ctrip Technology
Ctrip Technology
Efficient Online Performance Testing Using a Mirror Cluster Self‑Service Platform

Performance testing plays a crucial role in software and Internet services, affecting user experience, retention, and brand perception; therefore, achieving high‑efficiency online load testing is essential.

The goal addressed here is how to conduct online application load testing efficiently, emphasizing three efficiency characteristics: convenience (anyone can quickly start a test), safety (preventing production incidents), and simplicity (clear, traceable results).

The solution leverages a Mirror cluster whose core is container traffic forwarding, as illustrated in Figure 1. Mirror replay is chosen for online testing because it offers rich scenario coverage, easy server management, simple configuration, developer‑friendly operation, integration with offline self‑service platforms, and higher project support.

Mirror replay diagram
Mirror replay diagram

The technical background outlines Ctrip’s evolution from physical machines to virtualization and finally containerization, adopting sidecar patterns and container traffic mirroring (e.g., using pcap) to forward live traffic to sandboxed instances for scalable load testing.

Implementation focuses on three aspects:

Choosing a convenient and reliable request‑generation tool. The platform must handle many‑to‑many app‑service mappings, requiring whitelist verification to ensure correct routing.

Ensuring safety during online replay. This includes whitelist approval by service owners, automatic stop on error conditions detected via CAT monitoring, and preventing impact on production metrics.

Aggregating results and simplifying troubleshooting. Performance data from CAT/HickWall are consolidated for both application and server metrics, and tools such as JProfile, .NET dump analyzers, and Pass are provided for rapid issue analysis.

Current limitations include coarse‑grained replay scenarios, limited scaling for low‑traffic services, and potential data consumption errors when mirror services run without active replay.

In conclusion, the Mirror‑based self‑service performance testing platform is a key component of Ctrip’s overall automated testing ecosystem, complementing offline load‑testing tools, ES/Kafka replay utilities, and diagnostic tools to deliver a comprehensive, efficient, and safe performance testing solution.

Original Source

Signed-in readers can open the original source through BestHub's protected redirect.

Sign in to view source
Republication Notice

This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactadmin@besthub.devand we will review it promptly.

Operationsmirror clusteronline load testing
Ctrip Technology
Written by

Ctrip Technology

Official Ctrip Technology account, sharing and discussing growth.

0 followers
Reader feedback

How this landed with the community

Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.