The article argues that for typical file‑upload scenarios you should upload to MinIO first and then write to the database without trying to enforce strong distributed consistency, explaining the pitfalls of transaction‑based solutions and offering simple async cleanup strategies.
This article explores a series of unconventional yet practical caching designs—including consistent hashing with local caches, request‑scope caching, session‑level caching, client‑side caching, pre‑loading, and graceful degradation—to dramatically improve backend response times, reliability, and interview impact.
This article examines the common data‑consistency problems that arise when updating both a database and a cache under high concurrency, evaluates four typical write‑order strategies, and presents the most reliable solution—writing to the database first and then safely invalidating the cache using retry, scheduled tasks, MQ, or binlog listeners.
This article explains how unreliable networks, clock instability, and node failures can cause data inconsistency in distributed clusters, introduces the Raft consensus algorithm, details its roles, election process, log replication, read/write handling, consistency models, and mechanisms to avoid split-brain and livelock.
This article outlines practical cache design principles, covering suitable scenarios, health metrics, common pitfalls like avalanche, breakdown and penetration, and concrete implementation rules for both local (Caffeine) and Redis caches to ensure microsecond‑level response and stable high‑traffic performance.
PoleFS is a high‑performance, cloud‑native distributed file system that combines NVMe‑accelerated hot storage with S3‑based cold storage, offering multiple client access methods, multi‑level metadata and data caches, prefetch/warm‑up strategies, and a Close‑to‑Open consistency model to balance performance and data correctness.
This article analyzes the current bottlenecks of Bilibili's private messaging service, explains the technical challenges of massive data volume and traffic spikes, and presents a comprehensive multi‑layer architecture upgrade—including cache strategies, BFF refactoring, database sharding, and consistency mechanisms—to ensure scalability and reliability.
In the age of AI‑driven code generation, requirements must be both unambiguous and consistent to avoid misinterpretation, unpredictable outputs, and faulty logic, ensuring that generated code aligns with stakeholder intent and maintains quality.
This article explains MySQL master‑slave replication, why read/write splitting is essential, four practical read‑distribution strategies, the main causes of replication lag, four solutions for write‑after‑read consistency, multi‑slave allocation methods, replication modes, and a real‑world e‑commerce decision matrix, ending with a golden architecture recommendation.
This article explains why MySQL master‑slave replication can fall behind, outlines common causes of data inconsistency, and presents four comprehensive solutions—including architecture tweaks, multi‑threaded replication, configuration tuning, and Percona Toolkit verification—to achieve fast, stable, and accurate data synchronization.
Alibaba’s Taobao Starry Image Editing model tackles the e‑commerce challenge of maintaining visual consistency by introducing a high‑fidelity, plug‑in architecture, a million‑scale consistency dataset, and multi‑stage multilingual training, enabling precise, controllable edits without altering product layout or background.
This article systematically compares four cache update patterns—cache‑aside, asynchronous write‑back, read/write‑through, and ongoing optimizations—detailing their implementation steps, suitable scenarios, advantages, drawbacks, and practical tips such as delayed double deletion and proactive cache refreshing to balance performance and consistency.
This article explores nineteen core design patterns for distributed systems—including Bloom filters, consistent hashing, quorum, leader‑follower, heartbeat, fencing, WAL, segmented logs, high‑water mark, leases, gossip, Phi accrual detection, split‑brain handling, checksums, CAP and PACELC theorems, hinted handoff, read repair, and Merkle trees—explaining their purpose, operation, and typical use cases.
This article examines write conflicts in single-leader replication, comparing exclusive and shared data, exploring uniqueness constraints, async replication delays, and various conflict‑resolution techniques such as unique indexes, bitmap mapping, LWW ordering, and multi‑version control for collaborative editing.
This article explains the challenges of implementing causal consistency, compares it with linear and sequential consistency, describes how order guarantees are enforced in leader‑based replication, introduces Lamport timestamps and total‑order broadcast, and outlines practical approaches for achieving causal consistency in distributed systems.
This article explains how Redis and Zookeeper implement distributed locks, compares their performance, consistency, and reliability, and provides guidance on choosing the appropriate lock mechanism for different backend scenarios.
This article explores the purpose of database transactions, the various failure scenarios they guard against, the nuances of consistency across ACID, replica and CAP models, and how atomic commit and consensus mechanisms ensure reliable distributed data processing.
This article explores the concept of consistency in distributed systems, breaking down various consistency models—including linear, sequential, causal, and eventual—explaining their definitions, practical implications, and how they guide the design of high‑availability architectures and data replication strategies.
This article explains the principles of leader‑based and multi‑leader replication, compares synchronous and asynchronous modes, discusses consistency trade‑offs, conflict resolution strategies, and architectural variations for single‑ and multi‑data‑center deployments.
This article explores the principles of data replication, comparing shared memory, shared disk, and non‑shared storage architectures, detailing replication formats, consistency challenges, and various replication strategies such as synchronous, asynchronous, semi‑synchronous, and majority‑based approaches, helping engineers choose the right trade‑offs.
This article examines common cache‑database consistency challenges and compares four update strategies—including write‑through, cache‑aside, and delayed double‑delete—to help backend developers choose the most suitable approach for maintaining data integrity in high‑traffic systems.
Redis cluster relies on two ports per node, a hash‑slot based sharding scheme, a master‑slave replication model, and provides eventual consistency, making these four fundamentals essential knowledge for developers and operators to.
This article explains the difference between strong and weak consistency, explores Java's memory model, visibility and ordering issues, and shows how Java 8 redesigns ConcurrentHashMap with CAS and Unsafe to achieve strong consistency for put and get operations.
This article explains the CAP theorem and its practical extension BASE, describing their core concepts, trade‑off combinations, typical components such as Zookeeper, Eureka, and Nacos, and engineering techniques like asynchronous replication, Saga, and idempotent design for building highly available distributed systems.
This comprehensive guide explores the fundamentals of distributed systems, covering theoretical foundations, architecture design patterns, consistency models, scalability, deployment, operations, and practical engineering practices for building robust microservice‑based solutions.
The article explains the principles, implementation techniques, trade‑offs, and typical use cases of strong consistency and eventual consistency in distributed systems, helping architects decide which model best fits the performance, availability, and correctness requirements of their applications.
This article examines why many large‑scale systems avoid the traditional delayed double‑delete cache‑invalidation strategy, explains its critical drawbacks, and presents alternative lease‑based and version‑based approaches with Lua scripts and Java wrappers for Redis to achieve stronger consistency.
An in‑depth guide explains the CAP theorem’s three pillars—Consistency, Availability, Partition Tolerance—illustrates why only two can be achieved simultaneously, and shows real‑world trade‑offs across e‑commerce, finance, and social platforms, while introducing the complementary BASE model for practical system design.
This article explains how Apache Paimon guarantees data timeliness and consistency through snapshot files, two‑phase commit, and configurable expiration policies, and it outlines practical optimization and cleanup techniques for maintaining efficient storage and query performance.
This article explains what distributed architecture is, traces its historical development, details core technologies such as consensus protocols and load balancing, explores real‑world applications in e‑commerce, fintech and big data, discusses key challenges and mitigation strategies, and looks ahead to its role in AI, IoT and blockchain.
This article explains why cache consistency is critical for Meta, how the company measures and monitors consistency, the design of the Polaris system that detects and resolves stale cache entries, and provides a concrete Python‑style example illustrating the challenges and solutions.
The article examines the write‑after‑read consistency problem in multi‑active cross‑region systems, compares single‑write‑single‑read routing, quorum‑based multi‑write‑multi‑read, and read‑while‑copy methods, explains why primary‑secondary replication is preferred, and proposes a four‑step framework—scenario flagging, data marking, latency evaluation, and near‑by asynchronous replication—to meet WAR requirements efficiently.
The article examines how frequent inventory adjustments in JD retail supply‑chain can suffer from concurrency issues, explains why traditional locking may fail, analyzes a real‑world case of lock misuse, and presents code‑level, database‑level, and architectural solutions to guarantee data consistency.
The article explains why architectural consistency—across style, implementation solutions, and formal design—is essential for preventing decay in evolving enterprise systems, and provides concrete practices such as unified patterns, standardized tech stacks, documentation, code reviews, and toolchain standardization to maintain flexibility and maintainability.
This article walks through a real‑world cache consistency bug in a high‑traffic push service, explains cache penetration, breakdown, and avalanche, compares strong and eventual consistency models, and presents practical cache update and invalidation strategies to prevent data mismatches.
Zookeeper, the open‑source coordination service modeled after Google’s Chubby, offers strong consistency for distributed applications, but its CP design, synchronous master‑slave replication, log‑heavy state changes, and limited health checks lead many large tech firms to favor more available alternatives, though it still fits certain use cases.
This article details Uber's Docstore distributed database challenges and explains the design, architecture, and implementation of CacheFront—a transparent, high‑performance caching layer that reduces latency, improves scalability, and maintains strong consistency across microservices.
This article details the practical journey of building and scaling an Elasticsearch‑based search system that supports tens of millions to billions of items, covering architecture design, capacity planning, multi‑data‑center deployment, data synchronization via RocketMQ and Flink, and multi‑layer reconciliation to ensure consistency and high QPS.
The article examines a cache‑consistency failure in Tmall International’s high‑concurrency push system, explains classic cache problems and mitigation techniques, analyzes the delete‑then‑update bug that caused null‑plan errors, and evaluates four corrective strategies ranging from double‑write to delayed double‑delete.
This article explains the Paxos consensus algorithm, detailing its roles, three-phase execution process, key properties such as consistency, availability and fault tolerance, and showcases its practical applications in distributed databases, file systems, and coordination services.
The article reviews Amazon MemoryDB, a Redis‑based in‑memory database that uses a distributed transaction log to achieve strong consistency and durable persistence for high‑throughput, latency‑sensitive workloads, and compares its architecture and performance with open‑source Redis.
This article explains how to build a multi‑level caching system for microservice applications, covering client‑side HTTP caching, CDN and Nginx static‑resource caching, in‑process and distributed Redis caches, consistency challenges, and practical guidelines for when such a design is beneficial.
This article explains Meta's approach to cache invalidation and consistency, detailing why ultra‑high consistency matters, how their Polaris monitoring system detects and resolves inconsistencies, and provides a simplified Python example that illustrates the underlying mechanisms and challenges.
This article explains Meta's approach to cache invalidation and consistency, why ultra‑high consistency matters for user experience, the monitoring infrastructure they built, the Polaris system that detects and repairs inconsistencies, and provides a concrete Python‑style code example illustrating the problem and solution.
While both vectorization and hashing are essential for handling large datasets, this article examines whether vector embeddings can match the deterministic consistency of hash functions, comparing their collision handling, data structure design implications, and suitability for retrieval and machine‑learning tasks.
This article explains why read‑write separation is essential for high‑concurrency systems and details three MySQL master‑slave synchronization methods—half‑synchronous replication, database‑middleware synchronization, and cache‑key write tracking—along with their advantages, drawbacks, and consistency considerations.
This article revisits a 2017 decision to ban exceptions in a large codebase, examines the technical and organizational reasons behind that rule, analyzes exception mechanics, coroutine interactions, and performance impacts with concrete C++ examples, and finally proposes a pragmatic action plan for safely adopting exception‑driven design.
This article introduces eight essential distributed‑transaction techniques—including 2PC, 3PC, TCC, Saga, distributed locks, local message tables, reliable message transactions, and best‑effort notifications—detailing their workflows, advantages, drawbacks, and suitable business scenarios to help engineers choose the right solution.
The article explains how payment systems should link related documents to maintain status and amount consistency, implement idempotency through unique keys, and use progressive retry mechanisms with message queues to handle network glitches and ensure reliable transaction processing.
This article deep‑dives into Nacos as a service registry, explaining the differences between temporary and permanent instances, registration mechanisms across 1.x and 2.x versions, heartbeat and health‑check strategies, service discovery methods, data‑consistency models, and the underlying data model that powers Nacos clusters.
Learn how the CAP theorem defines the trade‑offs among consistency, availability, and partition tolerance, and see how different NoSQL databases such as MongoDB (CP) and Cassandra (AP) align with these principles to guide your choice of storage for cloud‑native, distributed applications.
This article explains why introducing a cache improves read performance as traffic grows, compares different cache‑database consistency approaches, analyzes their pros and cons in concurrent scenarios, and recommends using the "update database then delete cache" pattern with message queues or change‑log subscriptions to maintain data integrity.
This article compares the delayed double‑delete method with an asynchronous binlog‑subscription approach using Canal and RabbitMQ, explaining their workflows, pitfalls, and code implementations for maintaining cache and database consistency under high concurrency.
This article explains why a MySQL 5.7.30 upgrade with GTID enabled causes a GTID‑consistency error for CREATE TABLE ... SELECT statements, describes the underlying mechanism, provides a safe two‑step workaround, and notes that MySQL 8.0.21+ now supports atomic DDL for this operation.
This survey systematically reviews the research progress of in‑network storage systems, covering programmable network hardware characteristics, the two main challenges of building high‑performance solutions, task‑based classification of existing systems, representative designs, and future research directions such as switch‑NIC collaboration, multi‑tenant support, security, and automatic offloading.
The article details how to design and implement a cloud‑native payment fund account system on Tencent Cloud, covering account definitions, fund flow and multiple account types, TDSQL storage, separated fund and account services, robust security, distributed transactions, auditing, reconciliation, and high‑availability measures for high‑concurrency merchant payments.
This article explains how distributed systems handle business inconsistencies through compensation mechanisms, detailing rollback versus retry approaches, their implementation patterns such as explicit/implicit rollback, various retry strategies, and the trade‑offs between ACID and BASE consistency models for achieving eventual consistency.
To protect the outbox Redis cluster from extreme read amplification during hot events, the service adds a resident local cache for hot creators’ latest posts, using a threshold‑based list, change‑broadcast updates, and checksum verification, which achieved over 55% cache hits and cut peak Redis load by roughly 44% and CPU usage by 37%.
The article explains why testers need clear system boundaries and authoritative test criteria, outlines methods to identify system components, and presents seven consistency dimensions—user requirements, comparable products, historical behavior, visual design, statements, standards, and purpose—to ensure reliable and repeatable testing outcomes.
The article explains how distributed applications handle failures through business compensation mechanisms—defining rollback and retry strategies, consistency models, and practical considerations to maintain eventual consistency across multiple services.
This article explains the CAP theorem and its trade‑offs, introduces the BASE model as a practical compromise, and demonstrates how Zookeeper implements a CP registration center while Eureka adopts an AP approach, illustrating the impact on consistency, availability, and partition tolerance in distributed systems.
This article explains the fundamentals of cache systems, describes the evolution from no‑cache to distributed and local caches, analyzes common challenges such as consistency, hot‑key detection, and cache avalanche, and provides practical guidelines and real‑world lessons for designing effective backend caching solutions.
This article explains the concepts of sequential, causal, and eventual consistency in distributed systems, illustrating their guarantees with examples and diagrams, and discusses related consistency guarantees such as monotonic reads, writes, and read‑your‑writes, helping readers grasp how these models affect system behavior and performance.
This article explains consistency in distributed systems, distinguishing strong and eventual consistency, outlines the CAP and FLP impossibility theorems, and details common consensus mechanisms such as two‑phase commit, three‑phase commit, and the Paxos algorithm.
This article explains the fundamentals of distributed consistency, covering strong, weak, and eventual consistency, the CAP theorem, BASE model, and how ACID principles relate to modern distributed systems, helping engineers balance availability and data integrity in large‑scale architectures.
This article explains how distributed applications handle failures through business compensation mechanisms, detailing rollback and retry strategies, consistency models like ACID and BASE, and practical design considerations for microservice architectures.
This article examines various strategies for maintaining data consistency between caches (such as Redis and local memory) and databases during double-write operations, evaluates their advantages and drawbacks, and proposes solutions like delayed double deletion, message‑queue compensation, and binlog‑based cache updates.
This article explains the concept of consistency in distributed systems, distinguishes strong and weak (eventual) consistency, outlines typical use cases and challenges, and reviews key protocols such as 2‑Phase Commit, 3‑Phase Commit, Paxos, and Raft, while referencing the FLP and CAP theorems.
This article provides a comprehensive overview of database read‑write splitting, detailing its benefits, core technical implementations, practical best‑practice guidelines, and a comparison of popular open‑source and commercial products that enable efficient read‑write traffic distribution.
The article traces the evolution from Codd's relational model to modern RDBMS scaling limits, explains why centralized Hadoop/HBase architectures struggle with high‑concurrency workloads, and shows how Cassandra’s decentralized design—using consistent hashing, gossip, and virtual nodes—overcomes these bottlenecks while offering flexible consistency guarantees.
This article explains the principles, benefits, and challenges of master‑slave, multi‑master, and leaderless database replication, discusses replication lag, quorum consistency, conflict resolution, and presents various data partitioning techniques with practical examples and diagrams.
This article provides an in‑depth overview of Redis, covering its single‑threaded architecture, core data structures (String, Hash, List, Set, ZSet), persistence mechanisms (RDB, AOF, hybrid), replication, Sentinel and cluster designs, memory eviction policies, bitmap analytics, skiplist implementation, and strategies for ensuring data consistency.
The article examines why engineers favor reuse, outlines the logical arguments that link reuse to productivity, exposes common misconceptions and hidden costs, and proposes autonomy and consistency metrics as practical criteria for deciding when and how to abstract shared components in software systems.
The article examines common microservice transaction consistency techniques—including blocking retries, asynchronous queues, TCC compensation transactions, local message tables, and MQ transactions—explaining their mechanisms, advantages, drawbacks, and practical code examples for ensuring data integrity across distributed services.
The article examines the common belief that reusing code and components boosts productivity, explains the flawed reasoning behind it, highlights the hidden costs of excessive abstraction, and proposes autonomy and consistency metrics as guidelines for sensible system design.
The article explains various techniques for ensuring data consistency in distributed microservice architectures, including blocking retries, asynchronous queues, TCC compensation transactions, local message tables, and MQ transactions, while discussing their advantages, drawbacks, and practical implementation details.
This article explains fundamental distributed‑system mechanisms—including Bloom filters for space‑efficient membership tests, consistent hashing for scalable data placement, quorum requirements for operation safety, leader‑follower coordination, heartbeats, fencing, write‑ahead logging, log segmentation, high‑water marks, leases, gossip protocols, failure detection, split‑brain resolution, checksums, CAP and PACELC theorems, hinted handoff, read repair, and Merkle trees—providing a comprehensive overview for engineers.
The article explains how using a Redis cache can boost read performance but introduces consistency challenges, compares four cache‑aside write strategies, and recommends the reliable “update database then delete cache” approach combined with short expiration times and asynchronous message‑queue invalidation to keep data fresh.
This article examines cache update strategies in high‑concurrency systems, analyzing the trade‑offs of delete‑then‑write versus write‑then‑write approaches, illustrating race conditions, and presenting solutions such as CAS, distributed locks, asynchronous updates, and delayed double‑delete to maintain data consistency.
The article presents the architecture of WLock, a Paxos‑based distributed lock service, analyzes key isolation schemes, evaluates cluster expansion and splitting, and details a multi‑step key migration process—including forward and reverse migration, node scaling, and consistency safeguards—to achieve high‑availability and isolated lock handling in multi‑tenant environments.
The article analyzes the challenges of keeping cache and database consistent after write operations, evaluates four update‑order strategies, discusses their concurrency pitfalls, and presents practical solutions such as CAS, distributed locks, asynchronous updates, and delayed double‑delete techniques.
The article describes ShopeePay’s cloud‑native, high‑availability Naming Center built on sRPC, which replaces etcd with an in‑memory, peer‑to‑peer service registry that uses gossip heartbeats, self‑protection, and a coordinator for eventual consistency, rapid recovery, and seamless upgrades.
The article presents the architecture, business scenarios, consistency mechanisms, double‑entry accounting, hotspot‑account handling, storage sharding, and regulatory compliance of Baidu's transaction middle‑platform asset system that supports C2C payments, virtual‑coin, cash, and asset services across dozens of products.
The article examines the common belief that reuse improves productivity, explains the logical fallacies behind it, highlights the hidden costs of over‑generalization, and proposes autonomy and consistency metrics as practical guidelines for evaluating when reuse truly adds value.
This article explores how to achieve data consistency by examining transaction properties, isolation levels, common concurrency issues like dirty, non-repeatable, and phantom reads, and compares lock‑based concurrency control with multi‑version concurrency control (MVCC), detailing undo logs, read‑view mechanisms, and practical examples.
This article explains the design and implementation of Baidu's transaction middle‑platform asset system, covering C2C payment models, core business scenarios, asset flow processes, consistency guarantees, double‑entry accounting, hotspot account handling, storage architecture, and regulatory compliance to support rapid asset‑related business growth.
This article details the design and implementation of Baidu's transaction middle‑platform asset system, covering its C2C payment model, core asset management capabilities, consistency guarantees, double‑entry accounting, hot‑account handling, sharding and archiving strategies, and compliance with national financial regulations.
This article explains why caching is essential for high‑traffic services, describes the inherent consistency problems between Redis and MySQL, analyzes four cache‑update strategies with concrete concurrency scenarios, and provides practical recommendations such as using delete‑after‑update, delayed double‑delete, expiration, MQ, and binlog subscription to achieve eventual consistency.
This article explains essential distributed‑system design patterns such as Bloom filters, consistent hashing, quorum, leader‑follower replication, heartbeats, fencing, write‑ahead logs, segment logs, high‑water marks, leases, CAP and PACELC theorems, hinted handoff, read‑repair, Merkle trees, and related failure‑detection mechanisms, illustrating how they improve scalability, consistency and fault tolerance.
This article explains how transactions, consistency, and consensus interrelate in distributed systems, reviews replication models, details ACID properties, internal consistency models, consensus algorithms, and Kafka’s configuration and transaction support, and highlights verification tools like TLA+ and Jepsen for ensuring correctness.
The article surveys three data‑replication models—master‑slave, multi‑master, and leaderless—explains how they enable scalability and fault‑tolerance, and examines core distributed‑system challenges such as partial node failures, unreliable networks, and unsynchronized clocks, while stressing safety‑liveness trade‑offs and design techniques like quorum and timeouts.
This article presents a concise overview of essential distributed‑system design patterns—including Bloom filters, consistent hashing, quorum, leader/follower, heartbeat, fencing, high‑water mark, leases, gossip protocol, PACELC theorem, hinted handoff, read‑repair, and Merkle trees—explaining their purpose, operation, and typical usage in large‑scale storage and coordination services.
This article introduces essential distributed‑system mechanisms—including Bloom filters, consistent hashing, quorum, leader/follower roles, heartbeats, fencing, write‑ahead logs, segment logs, high‑water marks, leases, gossip protocols, failure detection, CAP/PACELC theorems, hinted handoff, read‑repair, and Merkle trees—to help engineers design scalable and fault‑tolerant services.
This article examines why cache consistency cannot be fully guaranteed, analyzes common inconsistency scenarios between Redis and MySQL, compares four update strategies, and offers practical recommendations such as updating the database before deleting cache, using short expiration times, delayed double‑delete, MQ or binlog approaches to achieve eventual consistency.
This article explains the design, six mechanisms, and source‑code flow of Nacos’s Distro protocol, showing how asynchronous replication, periodic sync, new‑node loading, and local reads together provide AP‑style high availability and eventual consistency for service registration.
The article explains the principles, requirements, and common implementations of distributed lock services using Redis and ZooKeeper, analyzes their drawbacks, and introduces SharkLock's design choices—including lock metadata, reliability via SharkStore, dead‑lock prevention, safety mechanisms, and Raft‑based consistency—to guide developers in building robust distributed locking solutions.
Meta’s engineering team describes how they raised cache consistency from six‑nines to ten‑nines by defining precise invalidation semantics, building the Polaris observability service, and implementing systematic tracking of cache mutations, offering practical strategies that apply to any distributed cache such as Redis or TAO.
The article examines why robust fault‑tolerance is essential for data stores, explains how transactions simplify error handling, breaks down the ACID guarantees, illustrates isolation and durability with concrete examples and diagrams, and discusses when multi‑object transactions are truly needed.
This article explains the challenges of distributed transactions in multi‑node systems, reviews common solutions such as 2PC, TCC, Saga, and focuses on implementing eventual consistency using message queues, detailing local message tables, MQ‑based transactions, and loss‑prevention techniques across RocketMQ, Kafka, and RabbitMQ.
Meta engineers describe the challenges of cache invalidation and consistency in large‑scale distributed systems, explain why stale caches are problematic, present their Polaris observability service and consistency‑tracking techniques, and detail how they raised TAO’s cache consistency from six‑nines to ten‑nines.
