Ctrip Architecture Evolution: Operations, Framework, and Application Layers

Ctrip's architecture has undergone continuous evolution and iteration, with many products experiencing more than five major updates. Each iteration addressed previous pain points while introducing new challenges, offering valuable insights for technical professionals.

Architecture Composition

The overall architecture consists of three parts: Operations, Framework, and Applications.

01 Operations

Operations provide the backbone for high availability and stability, featuring four key highlights:

Cluster management with SLB traffic control and automatic health‑check based scaling.

FullDR mechanism for Web, DB, and Redis clusters to ensure disaster recovery.

DBA strategy using M‑S and FullDR to guarantee data high‑availability and seamless migration from MSSQL to MySQL.

NOC 24/7 monitoring with order‑level dashboards and anomaly charts.

02 Framework

The framework layer has evolved through several milestones, including:

SOA & Gateway as a long‑standing service governance platform.

A proprietary release system with features such as braking, rollback, version switching, shared DLL packaging, and POM checks.

A custom message queue combining the strengths of open‑source solutions (Storm, MSMQ, ActiveMQ, RabbitMQ) with partitioned ordering, asynchronous compensation, and lifecycle tracking.

Configuration management progressing from simple web.config wrappers to JSON‑based, open‑source, listener‑enhanced systems.

03 Applications

Common application‑level techniques include PreLoading & LayerLoading, sharding, circuit breaking, rate limiting, and degradation, which together improve website and app stability, especially under adverse network conditions.

Architecture Evolution

Release System : Four generations—ITSM (C/S), CITSM (B/S), CRoller/ROP, and Tars (CD) – each addressing limitations of the previous version, culminating in remote backup and improved fault tolerance.

Configuration Management : Evolved from simple web.config editors to JSON‑based, open‑source systems with advanced listening mechanisms.

SOA : Progressed from an ESB‑based governance bus (first generation) to direct service connections (second generation), then to a feature‑rich third generation with circuit breaking, rate limiting, and dynamic routing, and finally to a Gateway that adds anti‑scraping and authentication for mobile/H5 services.

User Profile Project (Big Data)

The "User Profile" project serves as a core big‑data component, encompassing registration, collection, computation, storage, query, and monitoring. Data is ingested via batch and streaming pipelines (Kafka + Storm and Ctrip's Hermes), stored in Hive, MySQL, and Redis with FullDR + M‑S design, reaching over 100 billion records while maintaining ~10 ms average response time through sharding, circuit breaking, rate limiting, and degradation.

Overall, Ctrip's architecture demonstrates a systematic approach to scaling, reliability, and continuous improvement across operations, framework, and application layers.