Baidu DuoLiXiong Platform Stability Construction: Practices and Insights

This article introduces Baidu's DuoLiXiong local life services platform stability construction practices. DuoLiXiong is a SaaS solution for the local services industry, comprising three main products: Merchant Platform (for store management, order processing, after-sales, and fund withdrawals), Operations Platform (for merchant review, product review, and content management), and User Platform (for end users and influencers across Baidu and WeChat mini-programs).

The platform faces significant challenges due to its distributed microservices architecture: managing numerous microservices (user, product, order, merchant, coupon, payment), ensuring performance and stability across long call chains, maintaining data consistency with multiple external dependencies, and balancing rapid iteration with architectural health.

The stability construction approach focuses on three pillars: technical specifications, business specifications, and microservices practices. The implementation process covers: solution design (requirements analysis, technical architecture, interface design, storage design, compatibility, monitoring alerts, deployment planning), technical review (documentation review, admission rules, periodic reviews), coding standards (coding, security, MySQL, logging, exception handling), CodeReview (individual and centralized reviews), deployment (release windows, pre-deployment preparation, preview deployment, phased rollout), issue handling (immediate notification, damage control, root cause analysis), and case documentation.

Key technical practices include: eventual consistency using local message tables to handle asynchronous calls and ensure data integrity; idempotency design using deduplication tables to prevent duplicate submissions, timeout retries, message duplicate consumption, and high-concurrency issues; monitoring using Trace for full链路 tracing, Tianyan (one-stop log service platform) for log collection and alerting, and Prometheus/Grafana for metrics visualization.

Future plans include automated scaling based on custom Prometheus metrics for flash sales and promotions, and intelligent fault tolerance for core business processes (ordering, payment, verification) and dependent services (Redis, MQ) to ensure user experience.