Senior engineer at a top internet giant, sharing Java, AI, tech knowledge, growth insights, and interview experiences.
This article explains the three fundamental challenges of distributed locks—availability, deadlock, and split‑brain—compares database and Redis implementations, walks through the five evolutionary steps of Redis locking, and provides a structured interview answer framework to demonstrate deep understanding.
The article breaks down a complete flash‑sale system into four phases and explains how Redis distributed locks, CDN static pages, Nginx rate limiting, message‑queue peak shaving, and sharding together prevent overselling, crashes, and lost orders even when traffic surges dozens of times.
The article explains why reciting the 2PC protocol fails in real‑world interviews, then walks through the two dominant solutions—two‑phase commit and MQ‑based eventual consistency—detailing their mechanisms, trade‑offs, and a concrete interview response template that showcases practical experience and theoretical depth.
The article walks through five interview‑style steps—modulo hashing, consistent hashing, range sharding, metadata routing with etcd, and choosing Raft or Gossip—to design a highly scalable, hot‑item‑aware, and consistent distributed storage system for billions of products.
The article walks through Elasticsearch’s document retrieval workflow, covering single‑document GET, multi‑document mget, the two‑stage query_then_fetch mechanism, shard routing, load balancing, query and fetch phases, collectors, rescore, and pagination optimizations.
This article walks through Elasticsearch’s document indexing pipeline—from single‑document and bulk request routing, through the write‑refresh‑flush‑merge cycle, to the roles of Lucene, the translog, primary and replica shards, and how reliability, consistency, atomicity, isolation, real‑time visibility and performance are achieved.
This article explains Elasticsearch’s overall structure, the underlying Lucene index layout, the indexing and search processing flow, and how built‑in analyzers work, illustrating each concept with diagrams, token examples, and query scenarios.
This article dissects RocketMQ's concurrency mechanisms, explaining why it uses ReentrantReadWriteLock with HashMaps, how Semaphore controls async send rates, the SemaphoreReleaseOnlyOnce wrapper, the CountDownLatch‑based sync‑to‑async pattern, and the CompletableFuture redesign that reduces blocking in message replication.
