New concepts in the Agent space are arriving faster than most people can keep up with. Tool Calling, ReAct, RAG, Memory, Orchestrator — without understanding these terms, technical docs might as well be written in another language. Knowing the concepts but not the underlying mechanics leaves you stuck when real problems hit. This article breaks down each core concept from two angles: the product perspective (what user problem does this solve) and the technical perspective (how is this actually implemented). By the end, you’ll be able to discuss requirements with product managers and implementation details with engineers.
The Evolution of Agent Architecture: LangChain vs LangGraph
LangChain and LangGraph are the two most widely used frameworks in LLM application development. Developers often get confused about where each one fits: when should you reach for LangChain, when is LangGraph the right call, and can you use them together? This article breaks down the comparison across three dimensions: technical fundamentals, design motivation, and real-world usage.
Deep Dive into Kong API Gateway
Background
If you work with microservices, you can’t avoid an API gateway. Whether you use Spring Cloud Gateway, Envoy, or hand-roll something with Nginx, you need a unified entry point to route requests to backend services.
Kong is one of the most mature open-source solutions in this space — 40K+ stars on GitHub, backed by a commercial company, with a solid plugin ecosystem and wide adoption at scale. But many developers stop at “oh, it’s just a gateway” without digging into its architecture, plugin system, or configuration model. This article does a systematic breakdown across architecture, plugins, configuration, load balancing, and more.
Deep Dive into 12306's Backup Ticket Queue System Design
“I’m 50th in line, but someone 200th got their ticket first — how?” This question has puzzled countless ticket buyers. The answer lies in the underlying design of the 12306 backup system: seat reuse, per-interval independent queuing, and event-driven matching. This article breaks down this world-class high-concurrency system across five dimensions — not just what it does, but why it works this way.
Introduction to Oracle SQL*Loader and Its Applications
Oracle SQL*Loader is a utility in the Oracle database management system for loading large volumes of data from external files into an Oracle database. It’s fast, flexible, and handles a variety of formats and file types. This article covers the basics of SQL*Loader, how it works, and practical use cases.
Sharding — Cross-Database Paginated Queries
Zero Copy
Deep Dive into BigDecimal String Conversion
BigDecimal is the go-to class for high-precision arithmetic in Java. When it comes to converting a BigDecimal to a string, three methods are available: toString(), toPlainString(), and toEngineeringString(). Developers often pick one without fully understanding what each does, which leads to subtle bugs in production. This article walks through the source code of each method, explains when each one is appropriate, and covers the pitfalls that tend to catch people off guard.