This article explains the challenges of Raft member changes, compares the two‑phase Joint Consensus approach with single‑step changes, discusses correctness and availability pitfalls, and offers practical engineering guidelines for implementing safe and efficient cluster reconfiguration.
This guide explains the fundamentals of etcd, its Raft‑based architecture, cluster planning, secure certificate generation, installation steps, service configuration, and verification commands to deploy a reliable, SSL‑protected etcd cluster for service discovery and configuration management.
This article explains the evolution of backend service architectures, introduces the CAP and BASE theories, and provides detailed walkthroughs of the Paxos and Raft consensus algorithms, covering their roles, constraints, protocol steps, and practical considerations for building reliable distributed systems.
From CAP and BASE trade‑offs to the rigorous Paxos consensus and the more approachable Raft protocol, this article explains how modern distributed systems achieve consistency despite partitions, failures, and latency, detailing roles, phases, and safety guarantees that underpin reliable micro‑service architectures.
This article provides a comprehensive overview of the Raft consensus algorithm, detailing node roles (follower, candidate, leader), the leader election process, mechanisms ensuring cluster consistency, handling of node failures, and strategies for resolving split‑brain scenarios in distributed systems.
This article provides a concise yet comprehensive overview of etcd, covering its background, core features, common application scenarios such as key‑value storage, service discovery, messaging and distributed locking, and detailed step‑by‑step installation methods for various platforms.
This article explains the challenges of distributed systems such as node failures and network anomalies, then introduces the CAP theorem, BASE theory, two‑phase and three‑phase commit protocols, and details consensus algorithms including Paxos, Raft, ZAB, and Amazon Dynamo's NWR model, highlighting their trade‑offs and practical usage.
This article provides a detailed overview of Nacos' evolution from its 1.x architecture to the 2.x version, explaining the layered design, service discovery model, identified shortcomings of the older version, and the enhancements introduced with long‑connection support, new data models, and performance optimizations.
JCQ, JD Cloud’s self‑developed cloud‑native message queue, evolved from a monolithic 1.0 version to a 3.0 architecture that separates compute and storage, introduces multi‑Raft scaling, dynamic resource allocation, and high‑availability features, offering flexible, low‑maintenance messaging for high‑concurrency workloads.
This article explains the fundamentals of distributed consistency, covering the CAP theorem, BASE model, two‑phase and three‑phase commit protocols, Paxos and Raft consensus algorithms, Zookeeper’s ZAB protocol, and Amazon Dynamo’s NWR model with vector clocks for conflict resolution.
This article explains the leader election mechanisms used in big‑data systems—ZAB in Zookeeper, Raft’s role‑based election, their drawbacks such as split‑brain and ZooKeeper overload, and how Kafka’s controller‑based design solves these issues with efficient partition leader selection.
This article provides a comprehensive overview of etcd, covering its core concepts, common application scenarios, multiple installation methods, v3 architecture, command‑line client usage, and secure TLS deployment for cloud‑native distributed systems.
This article explains the principles and workflows of Two‑Phase Commit, Paxos, and Raft, detailing their roles, phases, failure handling, and how they achieve distributed consensus in fault‑tolerant systems, including the coordinator‑participant model, proposal numbering, leader election, heartbeat mechanisms, and log replication processes.
This article introduces EPaxos, a leaderless distributed consensus algorithm, explains its motivation from Paxos and Multi‑Paxos, describes its core concepts such as instance spaces, dependencies, and deterministic reordering, and discusses implementation challenges and practical considerations for engineers familiar with Paxos or Raft.
This article provides a comprehensive overview of TiDB's architecture, detailing the roles of TiKV Server, Placement Driver (PD), TiDB Server, and the TiSpark component, and explains how Raft ensures data consistency across the distributed database system.
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.
This article surveys major blockchain consensus mechanisms—including Byzantine Fault Tolerance, Proof of Work, Proof of Stake, Delegated Proof of Stake, Practical BFT, Paxos, and Raft—detailing their principles, workflows, advantages, and challenges, and illustrates how they underpin distributed ledger security and performance.
This article provides a comprehensive overview of the Raft consensus algorithm, detailing its strong‑leader model, node states, leader election process, log replication mechanics, consistency checks, and single‑node configuration changes, while illustrating each concept with diagrams and code examples.
This article explores the evolution of distributed consensus—from Paxos to Multi‑Paxos, Raft, and EPaxos—examining their mechanisms, understandability, efficiency, availability, and suitable scenarios, while providing comparative analysis and thought‑provoking questions for practitioners in modern cloud systems.
This article explains the fundamentals of distributed consistency, covering weak and strong consistency, the CAP theorem, ACID and BASE models, and detailed overviews of 2PC, 3PC, Paxos, Raft, Gossip, NWR, Quorum, and Lease mechanisms, highlighting their trade‑offs and practical use cases.
Nacos 1.3.0 introduces an embedded relational database, unified cluster management, an upgraded Raft consistency layer, security patches, Snowflake ID configuration, data migration guidance, new cluster addressing modes, and a set of Open‑API operations for Raft administration, all aimed at simplicity, performance, and high availability.
Distributed systems need globally unique, often monotonic IDs, and this article examines common ID generation strategies—Snowflake, database auto‑increment, segment allocation, multi‑master databases, and Raft‑based consensus—evaluating each for high availability and high performance, and highlighting trade‑offs and implementation details.
This article, the fourth in a series on building a Raft consensus module in Go, explains how to add persistent state, improve command delivery semantics, optimize AppendEntries handling, and handle crash tolerance, while providing concrete Go code examples and practical testing tips.
This article explains how to extend a Go implementation of the Raft consensus algorithm to handle client commands, replicate logs across the cluster, manage commit pipelines, and ensure election safety, while providing detailed code examples and discussion of underlying concepts.
This article, the second in a series on Raft, explains the election mechanism, server states, timers, RPC handling, and partition scenarios while providing complete Go code examples for the consensus module and its interactions.
Vivo’s feature‑storage platform, built on Nebula Graph’s Raft‑based, storage‑compute‑separated architecture and exposed via Redis‑compatible proxies, meets massive, low‑latency AI data demands while offering strong consistency, horizontal scalability, backup, active‑active replication, and a roadmap toward general‑purpose KV, cloud‑native integration, and advanced storage engines.
This comprehensive guide explains what etcd is, its Raft‑based architecture, key terminology, read/write flow, leader election, common scenarios such as service discovery and distributed locking, and provides step‑by‑step instructions for single‑node and multi‑node deployments with practical command examples.
This article provides a thorough introduction to the Raft consensus algorithm, explaining its purpose, core components such as state machine replication, log and consensus module, leader‑follower model, client interaction, fault‑tolerance considerations, the CAP trade‑off, and why Go is a suitable implementation language.
This article introduces the Raft distributed consensus algorithm, explains its core concepts such as replicated state machines, leader‑follower roles, client interaction, fault tolerance, the CAP trade‑off, and why Go is a suitable language for implementing Raft.
This article examines the pressures driving banks to redesign their core database systems, compares three distributed‑access approaches, explains the components of modern distributed databases, analyzes two‑phase and three‑phase commit issues, and evaluates consensus algorithms, CAP/BASE trade‑offs, GTM design, and point‑in‑time recovery.
This article provides a comprehensive technical overview of TiDB, covering its HTAP design, TiKV storage engine with RocksDB and Raft replication, the mapping of relational tables to key‑value pairs, MVCC implementation, transaction handling, and the PD scheduler that balances replicas, leaders, and hot spots across a distributed cluster.
This article explains the CAP theorem, illustrates the Byzantine Generals problem, and provides detailed overviews of PBFT, Paxos (including Multi‑Paxos), and Raft consensus algorithms, highlighting their phases, roles, and practical considerations for achieving consistency in distributed systems.
This article compares Zookeeper and etcd, detailing their origins, architecture, APIs, strengths, and weaknesses, to help you decide which distributed coordination system best matches your project's requirements and programming language preferences.
This article traces etcd’s evolution from its CoreOS origins, explains its Raft‑based distributed architecture, details its API groups, versioning and watch mechanisms, and showcases typical scenarios such as metadata storage, service discovery, leader election, and distributed coordination in cloud‑native environments.
This article examines Google's high‑availability distributed cron design, distills its core requirements and algorithms, and then presents a streamlined implementation for a startup using etcd and Raft, followed by a thoughtful discussion on whether early‑stage companies should adopt a middle‑platform strategy.
This article explains linearizability in distributed systems, compares Raft read strategies, and describes how CB‑SQL uses range leases and node leases to achieve strong consistency while reducing overhead, illustrating the trade‑offs and evolutionary design choices behind its consensus mechanisms.
This article reviews the evolution of etcd’s read‑write mechanisms, explains the “Fully Concurrent Read” feature introduced in etcd 3.4, and presents experimental results showing how it dramatically reduces expensive read latency and improves overall throughput in Kubernetes clusters.
The article describes how Tencent Financial Technology (FiT) built a globally largest, financial‑grade distributed MySQL cluster with multi‑active data‑center architecture, Raft‑based failover, priority‑driven switching, and automated backup to achieve 99.999% availability for massive payment workloads.
This article explains the fundamentals of distributed consistency by introducing the Raft consensus algorithm, covering its roles, leader election, log replication, handling of split votes, random timeouts, and various failure scenarios such as leader crashes and network partitions.
This article explains distributed data consistency concepts, the CAP theorem, various consistency levels, and provides detailed overviews of three major consensus algorithms—Paxos, ZAB, and Raft—including their mechanisms, roles, and practical applications such as in CB‑SQL.
This article explains how CB‑SQL improves distributed transaction performance by introducing a transaction record table, parallel prepare requests, one‑phase commit for single‑shard writes, latch‑based consistent reads, transaction pipelines, and a staged parallel‑commit mode, dramatically reducing latency and I/O overhead.
This article explains the fundamentals of distributed consistency protocols, covering Lamport's logical clocks, the basics and multi‑instance extensions of Paxos, Zookeeper's ZAB algorithm, and the Raft protocol, while comparing their designs, strengths, and practical engineering considerations.
This article explains the fundamentals of distributed consistency, introduces key terminology, and provides an in‑depth look at Raft's leader election, term handling, log replication, failure scenarios, and comparisons with Paxos, ZAB, PacificA, and Viewstamped Replication.
This article examines multiple fault‑tolerant transaction designs—RSM‑based KV, RSM‑based transactions, shared‑storage approaches, high‑availability KV layers, and single‑node engine extensions—comparing their replication strategies, lock handling, and performance trade‑offs while raising open questions about ordering and consistency.
This article introduces the foundational concepts and principles of distributed systems—including basic concepts, consistency models, CAP theorem, logical clocks, and advanced protocols like Paxos, Raft, and Zab—serving as the first part of a comprehensive Big Data mastery series.
The article explains how the CAP theorem forces trade‑offs between consistency, availability and partition tolerance, then surveys distributed commit protocols (2PC, 3PC) and consensus algorithms (Paxos, Raft, Zookeeper’s ZAB), and shows their practical use in systems such as ZanKV that combine Raft with RocksDB for strongly consistent, fault‑tolerant key‑value storage.
This article explains the inner workings of etcd, covering its Raft consensus implementation, node election and state management, as well as the persistent storage layer built on BoltDB, and includes detailed code excerpts to illustrate each component.
This article explains the principles, requirements, and implementation details of distributed lock services, comparing Redis and ZooKeeper approaches, and introduces SharkLock's design built on SharkStore with Raft-based replication, covering lock acquisition, release, reliability, scaling, and failover mechanisms.
This article explains the principles of distributed locks, compares Redis‑based and ZooKeeper‑based implementations, discusses their limitations, and introduces the SharkLock system built on SharkStore with Raft‑based replication, detailing its reliability, dead‑lock prevention, and safety mechanisms.
An in‑depth overview of the Raft consensus algorithm explains its server states, RPC mechanisms, leader election process, log replication workflow, safety guarantees, and how the protocol handles failures, illustrated with diagrams and practical examples.
This article details the design, architecture, and implementation of ZanKV—a high‑performance, distributed key‑value store that combines RocksDB storage, etcd‑Raft consensus, and a Redis‑compatible protocol, covering data partitioning, namespace isolation, expiration strategies, cross‑datacenter deployment, and performance tuning.
RadonDB is a next‑generation cloud‑native distributed relational database that combines the Raft consensus protocol with MySQL to deliver high availability, strong consistency, seamless scalability, and native support for OLTP, OLAP, distributed transactions, and comprehensive monitoring and backup features.
This article reviews RadonDB, a MySQL‑based distributed database, detailing its architecture, SQL, storage, and compute nodes, sharding strategy, data‑balancing mechanisms, migration process, and practical recommendations for DBAs seeking high‑availability and consistent relational data handling.
The article compares Raft, a crash‑only consensus used in private blockchains with leader election and O(N) communication, to PBFT, a Byzantine‑tolerant protocol for consortium chains that requires 3f+1 nodes, uses three‑phase messaging, O(N²) complexity, and different fault‑tolerance limits.
This article explains the fundamentals of Raft and PBFT consensus algorithms, compares their processes, fault tolerance limits, and suitability for private and consortium blockchains, and highlights key differences in leader election, message handling, and fault models.
This article presents an in‑depth overview of Zeppelin, a high‑availability, high‑performance block storage service, covering its motivation, online vs offline storage distinctions, data distribution strategies, centralized meta‑server design, replication policies, RocksDB‑based storage engine, Raft‑based consistency protocol, threading model, client request flow, and fault‑handling mechanisms.
Distributed systems rely on consensus to ensure consistent results, but achieving it faces fundamental challenges such as network unreliability, node failures, and trade‑offs captured by the CAP theorem, FLP impossibility, and various algorithms like Paxos, Raft, and Byzantine Fault Tolerance, each balancing consistency, availability, and safety.
This article explains the core challenges of achieving consistency in distributed systems, describes consensus algorithms such as Paxos and Raft, discusses theoretical limits like the FLP impossibility and CAP theorem, and shows how trade‑offs among consistency, availability, and partition tolerance shape practical system design.
Since its inception, Facebook has relied on MySQL to handle data from over two billion users, but recent shifts toward NoSQL have led to the development of Apollo—a layered storage system inspired by Paxos, Raft, RocksDB, and custom APIs, aiming to improve scalability, latency, and fault tolerance.
This article explains how we designed and implemented a scalable OAuth token storage and refresh system using Tarantool’s in‑memory database, Raft leader election, sharding across multiple data centers, and a custom lightweight queue to handle high‑throughput token updates while maintaining consistency and fault tolerance.
Tencent Cloud CMQ is a Raft‑based distributed message queue that delivers high reliability, strong consistency, and horizontal scalability for finance‑grade workloads, using multi‑node broker sets with majority‑acknowledged writes, automatic leader election, unlimited buffering, full‑path tracing, while requiring application‑level idempotency and offering limited strict ordering.
This article explains the theory and practical techniques behind synchronizing state across data centers using a two‑site three‑center model, covering distributed protocols like Paxos and Raft, CAP trade‑offs, latency challenges, and high‑availability designs for robust database replication.
TiDB’s founder Huang Dongxu recounts the journey of building a MySQL‑compatible, distributed database—detailing early challenges, architectural decisions, the extensive testing, the adoption of Go, Rust, Raft, RocksDB, and the emphasis on metrics, cloud‑native design, and open‑source community collaboration.
This article explains the fundamentals of consensus in distributed systems, compares Paxos and Raft, and provides a step‑by‑step guide with code snippets on embedding the etcd/raft library into the open‑source Mesos scheduler Swan to achieve reliable multi‑node data synchronization.
This article provides an in‑depth exploration of distributed consensus protocols, detailing the basics of Paxos, extending to Multi‑Paxos, and comparing it with Raft and Zab while discussing leader election, quorum, lease mechanisms, and practical considerations for implementing these algorithms in real‑world systems.
This talk walks through the evolution of databases, explains the challenges of building a distributed key‑value system that provides full ACID properties, and details the architectural components, sharding, migration, snapshot isolation, two‑phase commit, timestamp handling, and distributed transaction processing techniques.
This article introduces etcd, a highly‑available key‑value store based on the Raft algorithm, and explores its classic use cases such as service discovery, messaging, load balancing, distributed coordination, locks, queues, monitoring, leader election, and compares it with ZooKeeper, highlighting its simplicity, security, and cloud‑native advantages.
This article introduces the classic Paxos algorithm, explains why it is considered complex, and then presents the more approachable Raft consensus protocol, detailing its node roles, leader election, log replication process, and useful resources for deeper learning.
etcd, a highly‑available key‑value store built on the Raft algorithm, provides simple HTTP/JSON APIs for secure, fast, and reliable shared configuration and service discovery, enabling use cases such as service registration, load balancing, distributed locks, queues, leader election, and real‑time cluster monitoring.