Evolution of Multi-Level Caching Architecture: From XML to Protobuf with LocalCache

This article details the evolution of caching architecture in a rapidly growing business environment, where increasing data complexity and user traffic necessitated multiple iterations of caching solutions. The author describes four major phases of caching development:

Phase 1: XML Format Data Caching (1.0) - Initially used for simple C# applications to reduce database I/O pressure and improve response speeds during traffic peaks.

Phase 2: XML and .Net DataTable Format Data Caching (2.0) - Added Redis as a caching layer to support multiple servers, improve data consistency, and address performance issues with XML serialization/deserialization.

Phase 3: Protobuf Format Data Caching (3.0) - Adopted Protocol Buffers for cross-platform, cross-language support, offering superior performance with binary storage that's 3-10x smaller and 20-100x faster than XML.

Phase 4: Protobuf with LocalCache Multi-Level Caching (4.0) - Implemented a two-tier system with Redis and LocalCache to handle high-frequency cache reads, reduce network I/O overhead, and prevent server overload during traffic spikes.

The article includes detailed diagrams showing cache architecture evolution, data synchronization processes, and implementation code examples in Java and Go. Performance testing showed a 70% improvement in API response times. The author concludes by discussing future plans for real-time data updates to further enhance user experience.