This article examines the defining traits of large‑scale web systems and walks through the step‑by‑step evolution of their architecture—from single‑server LAMP setups to multi‑layer, distributed, and automated designs—highlighting performance, availability, scalability, and security considerations.
The article uses Starbucks' coffee‑ordering workflow as a real‑world analogy to explain asynchronous messaging patterns, correlation challenges, exception handling strategies, and why two‑phase commit can hinder scalability, illustrating how everyday processes inspire robust distributed system designs.
This article introduces the limitations of traditional local and distributed transactions for microservices, explains the BASE theory, and details four practical patterns—reliable event notification, maximum‑effort notification, business compensation, and TCC—providing code examples, diagrams, and a comparative table to guide developers in achieving eventual consistency across microservice architectures.
This article uses Taobao's backend as a case study to illustrate how a system evolves from a single‑machine deployment to a multi‑layer, highly available, cloud‑native architecture capable of handling millions of concurrent users, covering concepts such as distribution, load balancing, caching, database sharding, micro‑services, containerization, and cloud platforms.
Uber engineers rebuilt their payment system, Gulfstream, into a SOX‑compliant, job‑based double‑entry ledger that migrates billions of transactions across asynchronous services while maintaining zero‑downtime, high availability, idempotency, and strong data consistency.
Quorum, the core principle behind majority read/write and Paxos, can be defined in various ways—including weighted, hierarchical, and non‑majority quorums—to trade off system availability, latency, and fault tolerance, with examples illustrating how different quorum designs affect performance in distributed storage and coordination services.
This article outlines why distributed systems are needed, explains how they address cost and high‑availability issues by coordinating cheap nodes, and discusses the new coordination challenges such as service discovery, load balancing, fault isolation, monitoring, data partitioning, replication, and distributed transactions, providing a roadmap for further study.
This article outlines the evolution of project architecture from single‑server MVC to RPC, SOA, and micro‑services, and provides clear explanations of core distributed‑system terms such as clusters, load balancing, caching, and flow control for developers.
This article introduces RocketMQ’s core concepts, including its deployment architecture, naming server, broker and client roles, subscription model, consumption modes, queue allocation algorithms, rebalancing, offset storage, transaction and delayed messages, as well as filtering mechanisms, providing a solid foundation for further practice.
This article explains the evolution of project architectures—from single‑server MVC to RPC, SOA, and micro‑services—while defining key terms such as clusters, load balancing methods, and various caching strategies essential for high‑concurrency, large‑scale backend systems.
An experiment with a three‑node MySQL Group Replication (MGR) cluster shows that even when flow control is disabled, adding significant network latency to a single node reduces overall throughput, because MGR’s multi‑paxos protocol makes the delayed node a performance bottleneck during its turn as leader.
This article reviews various ID generation methods—including UUID/GUID, auto‑increment integers, random numbers, Snowflake, and MongoDB ObjectID—explains their advantages and drawbacks, and introduces the rpcxio/did distributed ID service with performance benchmarks and deployment considerations.
This article explains the open‑source Stateful Functions framework, its API and Flink‑based runtime, and how it simplifies building distributed stateful applications by combining serverless concepts with robust state management for event‑driven architectures.
This article explains how ClickHouse’s column‑oriented storage, vectorized execution engine, rich DBMS capabilities, flexible table engines, and carefully designed distributed architecture enable it to handle massive workloads with sub‑second query latency, making it a standout OLAP solution.
This article explains the concept of idempotency, examines common failure scenarios such as duplicate order creation, coupon redemption, and inventory deduction, and presents practical solutions—including unique identifiers, locking, database constraints, and state‑machine approaches—accompanied by concrete SQL and pseudo‑code examples.
This article walks through the step‑by‑step evolution of a high‑concurrency backend—from a single‑machine setup to a cloud‑native, micro‑service architecture—highlighting the technical challenges, key technologies, and design principles at each stage.
This article explains the motivations for adopting Spring Cloud, defines its components and design goals, compares it with Spring Boot and other frameworks, outlines its main sub‑projects, versioning scheme, and provides practical guidance on configuration, service discovery, load balancing, fault tolerance, and gateway development for cloud‑native microservices.
This article analyses the complete flash‑sale business flow, enumerates its unique characteristics and technical challenges such as high concurrency, bandwidth pressure, inventory overselling, and then presents a layered architecture—including frontend static pages, site‑level throttling, service‑level queuing, database sharding, caching, and anti‑cheat mechanisms—along with concrete Java code examples and best‑practice recommendations for building a reliable, high‑performance flash‑sale system.
This article explains the fundamental concepts of software architecture, distinguishes systems, subsystems, modules and components, describes various architectural layers such as business, application, data, code, technical and deployment, outlines evolution from monoliths to micro‑services, and highlights common misconceptions and measurement criteria for a sound architecture.
This article explains how Spring Cloud Gateway uses a token‑bucket algorithm backed by Redis and a Lua script to perform distributed rate limiting, reviews common limiting algorithms, provides detailed Java and Lua code examples, and analyzes each step of the implementation for high‑concurrency systems.
This article systematically examines the core technologies of OLAP systems, covering storage models, columnar formats, indexing, distributed storage architectures, query execution steps, optimizer designs, and emerging trends such as real‑time analytics, HTAP, cloud‑native deployment, and hardware acceleration.
This article explains the fundamental architecture of message middleware, its key roles such as peak shaving, asynchronous processing and decoupling, the two consumption models (publish‑subscribe and point‑to‑point), and introduces core Kafka concepts with practical Java code examples.
This article introduces the open‑source WPaxos project, explains the core Basic Paxos algorithm, analyzes various failure scenarios, and details the production‑level engineering optimizations and Java code implementations that enable high‑performance, reliable distributed consensus.
This article explains the fundamentals of distributed systems, covering the CAP and BASE theorems, caching strategies, message queues, database choices, JVM optimization, and practical architectural improvements applied to New Oriental's mobile app to enhance availability and performance.
This article systematically explains the fundamental concepts and typical pitfalls of distributed systems—including CAP and BASE theories, message‑queue reliability issues, distributed cache challenges, sharding strategies, and transaction models—while offering practical mitigation techniques for each problem.
Facing scalability limits, disaster risks, and latency in single‑region setups, OPPO built the JinS data‑sync framework, detailing its multi‑region challenges, design principles, modular architecture, consistency models, crash‑safe 2PC mechanisms, file‑queue optimizations, relay‑log handling, and performance gains over open‑source and commercial alternatives.
JD Digits' JACOBI Blockchain Innovation Lab unveiled a series of breakthrough research results—such as the Dumbo and Dumbo‑MVBA asynchronous consensus protocols, a blockchain identity management scheme, and a storage‑time proof—demonstrating leading academic contributions across consensus, cryptography, and distributed applications.
This article explains middleware as foundational software for distributed systems, outlines its evolution from early transaction monitors to modern Java and .NET stacks, categorizes its various types, and analyzes global and Chinese market sizes while profiling leading domestic vendors.
This article outlines why distributed systems are needed, explains how they address cost and high‑availability issues through coordinated nodes, and discusses the new challenges such as service discovery, load balancing, avalanche prevention, monitoring, data sharding, replication, and distributed transactions, while offering practical and theoretical learning paths.
This article provides a comprehensive overview of Apache RocketMQ, covering its core concepts, architecture, various message types, reasons for adoption such as asynchronous decoupling and peak‑shaving, as well as best practices and local transaction patterns for reliable distributed messaging.
This article examines Eastern Securities' transition to a microservice architecture by detailing the design, implementation, and performance of its gRPC‑Nebula service‑governance framework and the Star‑Chen platform, covering challenges such as heterogeneous interfaces, service registration, load balancing, fault tolerance, traffic control, multi‑registry support, and real‑world deployment results.
This article explains middleware fundamentals, its evolution in distributed systems, classifies its major types, analyzes global and Chinese market growth, and profiles key domestic vendors such as Dongfangtong, Baoland, Puyuan, Kingdee, Zhongchuang, Alibaba, and Tencent.
This article explains when Elasticsearch triggers a master election, describes each election stage—including active master and candidate selection, Bully algorithm comparison, and master node responsibilities—while providing code excerpts that illustrate the underlying implementation details.
This article explains the CAP theorem, its trade‑offs, the BASE model, various consistency levels, and how parameters N, W, and R are used to achieve strong or eventual consistency in distributed database systems.
This article explains etcd’s Raft‑based consensus, why odd‑numbered nodes are recommended, details the leader election process with log excerpts, discusses split‑brain and consistency guarantees, and provides step‑by‑step instructions for generating certificates, deploying an etcd cluster, and using etcdctl commands.
Tencent Cloud Block Storage (CBS) has evolved through three generations—apllo, atlas, and HiSTOR—adopting a client‑direct, distributed architecture, SPDK, RDMA and user‑space TCP to cut latency to sub‑microseconds while delivering exabyte‑scale throughput, high IOPS, and reliable multi‑copy replication for cloud VM workloads.
The article explains how distributed cache‑aside patterns work, why deleting stale cache entries is often preferable to updating them, analyzes basic and complex cache‑database inconsistency scenarios, and proposes a JVM‑queue‑driven, single‑threaded update mechanism with practical considerations for high‑concurrency environments.
This article explores the fundamentals of distributed systems, explaining what they are, why they’re used, design considerations such as replication and partitioning, the implications of the CAP theorem, common distribution strategies, typical architectural patterns, and the advantages and challenges of building and operating such systems.
This article explains why idempotency is essential, presents simple database‑based and concurrency‑safe implementations, and then details a generic, annotation‑driven idempotent framework with multi‑level storage, code examples, and deployment guidelines for Java backend services.
The iQIYI FinTech team implemented a Chaos Engineering framework, using a purpose‑driven Chaos Monkey to inject controlled failures, validate high‑availability, isolation, and self‑healing of payment services, derive architectural improvements, build a fault‑case library, and transition from fault detection to proactive system robustness.
This article analyzes a real incident where duplicate order numbers were generated under high concurrency, critiques the original timestamp‑based approach, and presents a thread‑safe Java solution using AtomicInteger, Java 8 date formatting, and optional IP suffixes to ensure globally unique identifiers.
This article explains the principles of primary keys, compares auto‑increment IDs and GUIDs, introduces the Snowflake ID structure, discusses its benefits and drawbacks, and provides a complete .NET implementation using Redis for dynamic worker‑ID allocation, including handling clock rollback and code examples.
The article explains the classic Cache Aside pattern, why deleting rather than updating cache is preferred, analyzes basic and complex cache inconsistency scenarios, and proposes a queue‑based lazy update solution with practical considerations for read‑write blocking, request routing, and hotspot handling in high‑traffic systems.
This article explains how Kafka consumer parameters like session.timeout.ms, heartbeat.interval.ms, and max.poll.interval.ms affect group rebalance, consumer liveness, and real‑time alert processing in high‑availability microservice architectures.
This article provides a comprehensive overview of RabbitMQ, covering its origins, core components, message routing patterns, persistence mechanisms, delivery modes, RPC support, clustering design, mirrored‑queue architecture, and flow‑control strategies for building reliable backend systems.
This article explains how Redis Sentinel provides automatic failover by combining multi‑replica deployment, health monitoring, leader election, and distributed consensus to detect master failures, select a new master, promote it, and notify clients, ensuring high availability for Redis databases.
This article describes how Zhihu's Moneta service, which stores over a trillion rows of user‑read data, migrated from MySQL sharding to the distributed NewSQL database TiDB to achieve high availability, horizontal scalability, millisecond‑level query latency, and improved overall system performance.
This article compiles common Redis interview questions, compares Redis with Memcached, explains eviction policies, and provides a concise introduction to distributed systems, high‑availability architectures, master‑slave replication, synchronization methods, and Redis‑based distributed lock solutions.
This article explains the fundamental concepts of software architecture, distinguishes system, subsystem, module, component, and framework, outlines architectural layers and classifications, describes the evolution from monolithic to distributed and micro‑service architectures, and discusses how to evaluate and avoid common design pitfalls.
This article reviews the award‑winning 2018 OSDI paper LegoOS, describing its split‑kernel architecture, component‑based resource management across processors, memory and storage, and how it enables hardware disaggregation in data‑center clusters while addressing network latency and failure handling.
Microservice architectures struggle with traditional ACID transactions, so this article reviews local and distributed transaction basics, explains why 2PC/3PC are unsuitable, introduces the BASE model, and details four practical consistency patterns—reliable event, async event, business compensation, and TCC—highlighting their mechanisms, advantages, and drawbacks.
The article analyzes a real‑world incident where duplicate order IDs were generated under high concurrency, demonstrates the shortcomings of the original Java implementation, and presents a thread‑safe redesign using AtomicInteger, Java 8 date‑time API and container IP suffix to guarantee unique identifiers across parallel requests and clustered instances.
The article explains, through interview excerpts, why many game studios avoid microservice architectures for their real‑time servers, highlighting latency‑sensitive communication, stateful processing, and the overhead of distributed networking that conflict with the performance demands of modern multiplayer games.
The article explains the fundamentals of distributed architecture, its high‑availability mechanisms, scalability benefits, common pitfalls such as data consistency, and real‑world case studies ranging from e‑commerce platforms to Nginx reverse proxy and microservice designs.
The article examines a real‑world incident where duplicate order IDs appeared under high concurrency, analyzes the shortcomings of the original timestamp‑and‑random‑based scheme, and presents a revised Java implementation using thread‑safe counters, Java 8 date‑time APIs, and IP‑based suffixes to reliably produce unique order numbers even in clustered environments.
This article provides a comprehensive overview of Apache HBase’s architecture, detailing its core components such as RegionServer, HMaster, ZooKeeper, WAL, MemStore, and HFiles, and explains key processes including read/write paths, compaction, region splitting, load balancing, and recovery mechanisms.
This article explains Kafka's role as a message system, details its fundamental components such as topics, partitions, producers, consumers, and replicas, describes how Zookeeper coordinates the cluster, and explores performance optimizations like sequential writes, zero‑copy, and network design.
This article explains why short URLs are used in spam SMS, outlines their benefits, describes the basic principle of mapping long URLs to short ones, and details service design choices such as storage, one‑to‑one mapping, high‑concurrency handling, distributed ID generation, and provides a Java implementation using Redis.
Sentinel 1.8.0 introduces extensive flow‑control, circuit‑breaker, and degradation improvements for microservices, adds native support for Java EE, Quarkus, and HTTP clients, and resolves critical bugs, making it a pivotal release for high‑availability distributed systems.
This article explains how to use Apache RocketMQ's 2PC-based transactional messaging feature, covering the overall workflow, key concepts such as half messages and compensation, and providing complete Java/Spring Boot code examples for producers, transaction listeners, and consumers.
This comprehensive article explains the core concepts of distributed systems—including node modeling, failure types, replica strategies, consistency levels, performance metrics, data distribution techniques, lease mechanisms, quorum, logging, two‑phase commit, MVCC, Paxos, and the CAP theorem—providing a solid foundation for designing robust, scalable architectures.
This article examines why many game companies avoid microservice architectures, highlighting the real‑time latency, stateful memory requirements, and networking constraints that make traditional microservice patterns unsuitable for fast, multiplayer game servers.
This article explains the classic Cache‑Aside pattern, analyzes common cache‑database consistency problems in high‑concurrency scenarios, and presents a lazy‑update queue solution that deletes stale cache entries, routes updates through internal JVM queues, and mitigates read‑blocking and hotspot issues.
This article outlines ByteDance’s adoption of chaos engineering, describing its background, industry examples, the evolution of internal fault‑injection platforms across three generations, the fault model and center design, experiment principles, and future plans for infrastructure‑level chaos and automated diagnostics.
ClickHouse, the high‑performance columnar OLAP DBMS behind Yandex.Metrica, combines complete DBMS capabilities, column‑oriented storage with compression, vectorized execution, flexible table engines, multi‑master clustering, and extensive SQL support, offering fast online queries and scalable distributed processing for massive data workloads.
Tencent’s TDMQ is a cloud‑native, Pulsar‑based message queue that separates storage and compute, offering multi‑protocol support, strong consistency, high reliability, active‑active cross‑region replication, and read‑only broker scaling to meet financial‑grade billing workloads with billions of daily transactions.
The article details how New Oriental's Cloud Classroom backend, built with Java, Spring, MySQL, Redis, Kafka, Sentinel, and other modern technologies, scaled to support millions of users and a hundred‑fold surge in demand during the pandemic through architectural optimizations, distributed caching, traffic control, and rapid performance improvements.
This article explains what ZooKeeper is, why it was created, its key features such as high performance, high availability, and strong consistency, and how it simplifies distributed application development by providing coordination primitives like naming, locks, leader election, and configuration management.
This article provides a comprehensive guide to Kafka's producer API, covering core concepts, client‑side workflow, essential configurations, idempotent and transactional producers, and practical Java code examples to help developers avoid common pitfalls and optimize message publishing.
This article explains how to handle session sharing in distributed systems through replication, sticky sessions, or a dedicated session server, and addresses cache penetration and avalanche issues with practical mitigation techniques such as request filtering, empty-object caching, hash repositories, Redis clustering, staggered expirations, and concurrency controls.
This article explains how Ant Financial’s logical data center (LDC) and unitized architecture, combined with OceanBase’s Paxos‑based consensus, enable the massive TPS growth for Double‑11 payments while addressing sharding, CAP trade‑offs, traffic diversion, and multi‑site disaster recovery.
The article describes the challenges of troubleshooting distributed deployments across many machines and presents a solution built on the ELK stack that centralizes logs from Java and Go services, enabling near‑real‑time search, visualization, and faster issue resolution.
This article traces the birth of middleware, the creation and multi‑generation evolution of SOFAStack at Ant Group, its adoption by banks and insurers, and its shift toward cloud‑native and trustworthy‑native architectures that power large‑scale financial services.
Apache Kafka 2.3 introduced static members and incremental cooperative rebalancing to reduce disruptive global rebalances, allowing workers to retain assignments during failures, schedule delayed rebalances, and improve scalability for Kafka Connect clusters, balancing availability and fault tolerance.
This article provides an in‑depth overview of Apache Kafka’s internal topics, preferred replicas, partition allocation mechanisms, log directory structure, index files, offset and timestamp lookup, log retention and compaction policies, storage architecture, delayed operations, controller role, consumer rebalance process, and producer idempotence.
The article surveys the Apache big‑data ecosystem, covering Hadoop’s storage and resource management, column stores HBase and Kudu, compute engines Spark, Flink, Impala, and Presto, coordination via ZooKeeper, ingestion with Sqoop and Flume, messaging Kafka, security Ranger and Sentry, metadata Atlas, OLAP Kylin, Hive, quality tool Griffin, notebooks Zeppelin, visualizations Superset and Tableau, the TPCx‑BB benchmark, and ends with an Alibaba analysis competition notice.
The article describes how a distributed deployment platform built a centralized Elasticsearch‑based log collection and query system to replace manual multi‑machine log inspection, detailing the background challenges, architecture, implementation steps, practical usage, and future improvements.
The article explains Kafka's reliability mechanisms, detailing how committed messages are persisted, common producer and consumer data‑loss scenarios, best‑practice configurations for acks, retries, replication, and offset handling, and describes idempotent and transactional producer setups for atomic writes.
This article explains Kafka's fundamental architecture, including the roles of producers, brokers, and consumers, key concepts such as topics, partitions, replicas, ISR, and controller, as well as detailed mechanisms of producer client structure, interceptors, serializers, partitioners, and consumer group rebalancing strategies.
This article surveys the rapidly evolving big data landscape by reviewing a wide range of Apache projects—including Hadoop, Spark, Flink, HBase, Kudu, Impala, Kafka, and others—detailing their core components, architectures, strengths, and typical use‑cases for building distributed data platforms.
This article details Beike's journey of storing and querying a 480‑billion‑triple industry graph in milliseconds, covering graph database fundamentals, a comparative evaluation of JanusGraph and Dgraph, the design and deployment of a Docker‑K8s based Dgraph platform, data ingestion pipelines, a custom Graph‑SQL layer, performance testing, optimizations, and future roadmap.
This article explains the fundamentals of hash functions, outlines the key characteristics of a good hash algorithm, and dives deep into consistent hashing—its background, mechanism, desirable properties, fault tolerance, scalability, and the use of virtual nodes to solve data skew in distributed systems.
This article provides a comprehensive technical overview of ClickHouse, covering its complete DBMS capabilities, column‑oriented storage and compression, vectorized execution engine, relational SQL support, diverse table engines, multi‑master clustering, sharding, and the design philosophies that make it exceptionally fast for large‑scale analytical workloads.
This article explains Kafka's replication process, covering producer acknowledgments, replica synchronization strategies, the roles of ISR and AR, the meanings of HW, LEO, LSO, LW, different ack levels, and how failures are handled to balance reliability and performance.
This article introduces Kafka Connect, a framework for building scalable data pipelines between Kafka and other systems, covering its architecture, key concepts like connectors and tasks, and practical deployment examples.
This article analyses the limitations of existing global unique ID solutions such as Snowflake, presents the design of a new high‑performance, time‑independent Mist algorithm with extended lifespan, and describes its practical Redis‑backed service Medis, including performance benchmarks and architectural considerations.
This article provides a comprehensive overview of Apache Kafka, explaining its distributed, fault‑tolerant architecture, horizontal scalability, disk‑based commit log, replication mechanisms, Streams API, KSQL, and why it is widely adopted as the backbone of event‑driven, high‑throughput systems.
The article explains how Alipay achieves massive payment throughput during Double‑11 by using logical data centers (LDC), unit‑based system design, multi‑active disaster‑recovery, and CAP‑theorem analysis, highlighting the role of OceanBase and PAXOS in ensuring consistency and availability.
The article walks through Taobao's backend architecture evolution—from a single‑machine setup to distributed caching, load balancing, database sharding, microservices, containerization, and finally cloud deployment—explaining each stage's technologies, challenges, and design principles for building scalable, highly available systems.
Consistent hashing, introduced by MIT in 1997, addresses hotspot issues in distributed systems by ensuring balance, monotonicity, spread, and load properties, using a ring hash space, virtual nodes, and minimal data movement when nodes are added or removed.
This article presents Ant Group’s financial‑grade unitized architecture, detailing industry‑standard distributed models, the design of RZone/GZone/CZone, its disaster‑recovery, elasticity and gray‑release capabilities, and real‑world banking case studies demonstrating cloud‑native deployment strategies in the financial sector.
This article explains the Cache Aside pattern, why deleting stale cache entries is often better than updating them, and presents a queue‑based lazy‑update solution that handles simple and complex consistency problems in high‑concurrency environments while outlining practical performance considerations.
This article explains the key characteristics of distributed systems, introduces the CAP and BASE theories, compares strong, weak and eventual consistency models, and reviews common distributed transaction solutions such as two‑phase commit, TCC and message‑based approaches, highlighting their trade‑offs and practical considerations.
This article explains the four main software architecture styles—monolithic, distributed, microservices, and serverless—detailing their structures, advantages, disadvantages, and typical use cases to help developers select the most suitable architecture for their projects.
The article introduces the JdHotkey framework, a lightweight, real‑time hotkey detection solution for Redis and other services, detailing the risks of hot keys, prior mitigation methods, core design components, operational workflow, and performance benchmarks across various hardware configurations.
This article explains the concept of microservices, contrasts it with monolithic architecture, outlines the drawbacks of monoliths, describes the advantages of microservice architectures, and briefly compares microservices with SOA, while also promoting a Java interview question resource.
This article explains the concept of high traffic, compares common mitigation techniques such as caching, degradation and rate limiting, and then details four classic rate‑limiting algorithms—counter, sliding window, leaky bucket and token bucket—followed by a practical Guava RateLimiter example and a brief note on distributed scenarios.
This article explains a lightweight transactional message solution for microservices that leverages RabbitMQ, MySQL, and Spring Boot to achieve eventual consistency, detailing design principles, compensation mechanisms, table schemas, and deployment considerations for high‑throughput asynchronous processing.
This article explains a lightweight, low‑intrusive transactional message solution for microservices using Spring Boot, RabbitMQ, and MySQL, covering design principles, table schemas, transaction synchronization, compensation mechanisms, code implementation, and deployment considerations to achieve eventual consistency without sacrificing performance.
This article explains the purpose of flow control, compares various rate‑limiting algorithms—including simple window, sliding window, leaky bucket, token bucket, and sliding log—provides Java interface definitions and code examples, discusses their complexity, precision, smoothness, and suitability for single‑machine and distributed scenarios, and offers practical deployment tips using Sentinel, Nginx, Guava, Tair, and Redis.
