This article explains how to build a real‑time typeahead (autocomplete) service that returns the most frequent query suggestions using an in‑memory Trie, sharding, offline aggregation, and caching to achieve sub‑20 ms latency, fault tolerance, and horizontal scalability.
REDtao is a high‑performance graph storage system built for Xiaohongshu that extends Facebook’s Tao architecture with a three‑layer hash structure, decoupled caching, leader‑follower distribution and cross‑cloud availability, delivering over 90% cache hits, 70% MySQL CPU reduction, 150 M QPS on a 16‑core server and seamless migration despite a 250% surge in daily‑active‑user traffic.
This article explains how to build a high‑concurrency ticket‑seckill system that can handle one million users buying ten thousand tickets simultaneously by using distributed load‑balancing, Nginx weighted round‑robin, Go‑based HTTP services, Redis atomic stock deduction, and practical performance testing.
Shopee’s off‑platform ads delay service combines Redis Zsets for expiration tracking, HBase for payload storage, and Kafka for queuing to reliably process up to 6 million tasks per minute with minute‑level delays ranging from one minute to thirty days, achieving horizontal scalability, fault tolerance, and a 75 % reduction in Kubernetes resource usage.
Baidu's Comment Middle Platform evolved from a single service into a robust middleware that delivers stable, high‑performance comment functionality across more than twenty products, handling hundreds of millions of daily requests with 99.995% SLA through graph‑based scheduling, tiered caching, and scalable sorting mechanisms.
This article details the architecture, data flow, module responsibilities, and successive optimizations of Baidu's short‑video push system, covering personalized timing estimation, user‑group management, frequency‑control redesign, and protobuf‑based compression to handle billion‑scale traffic efficiently.
Baidu’s Short Video Push System is a distributed platform serving hundreds of millions of users across multiple apps, delivering personalized, real‑time notifications via a modular architecture that includes material and user centers, recall, preprocessing, and delivery services, while optimizations such as activity‑based scheduling, bitmap‑based user segmentation, consistent‑hash frequency control, and protobuf compression boost click‑through rates, scalability, and resource efficiency.
The article describes the concept, architecture, and Java implementation of a distributed request amplifier in the DCS_FunTester framework that multiplies local QPS to achieve high‑throughput performance testing without affecting the tested service's load.
Short URLs are popular in SMS and social media because they save characters, look tidy, enable analytics, and hide parameters; this article explains their benefits, the basic redirect workflow, and provides a detailed backend design—including storage choices, high‑concurrency strategies, distributed ID generation, and a Java‑Redis implementation.
This article explores the architecture and implementation of a high‑concurrency ticket‑spike system, covering distributed load‑balancing, Nginx weighted round‑robin configuration, Go‑based local and remote stock deduction with Redis, performance testing, and strategies to avoid overselling and underselling.
This article describes the challenges of delivering large video lessons and other digital assets to distributed school campuses and explains how the Dandelion release platform uses resumable uploads, cloud OSS, long‑connection messaging, redundancy, and version‑controlled rollout to ensure reliable, low‑latency playback and scalable resource distribution.
This article explains how to design and implement a Redis‑driven, second‑level scheduled‑task cluster that reliably pushes millions of coupon notifications in real time, addressing latency, throughput, and single‑point‑failure challenges.
Youzan redesigned its ID generator to run independent March instances per room, sharing a single etcd cluster but allocating distinct, non‑overlapping ID ranges via boundary, lower, and upper parameters, enabling room‑level high availability, horizontal scaling, and zero‑downtime migration with automated tooling.
The article presents a comprehensive overview of the 58DP big data platform's task scheduling system, detailing its background, architecture, high‑availability design, slot‑based resource management, scheduling models, task lifecycle, priority rules, dependency handling, failure recovery, and future enhancements.
Elasticsearch is a distributed, open‑source search and analytics engine built on Apache Lucene, offering a simple REST API, scalability, and a rich ecosystem—including Kibana, Beats, and Logstash—that enables storage, querying, and aggregation of text, numeric, and geospatial data for diverse use cases.
This article explores the design of a high‑concurrency ticket‑spike system that can handle millions of simultaneous requests by combining multi‑layer load balancing, Nginx weighted round‑robin, Redis for atomic stock control, and Go‑based services, complete with code samples and performance testing.
TurboSearch is Tencent AI Lab's next-generation large-scale search system, delivering distributed massive indexing, high-performance parallel retrieval, multi-granularity and multi-modal vector indexing, private Docker deployment, integrated NLP query analysis, extensible plugins, and robust operations for massive data and diverse search scenarios.
This article presents a distributed system that efficiently supports large‑scale n‑gram language models for automatic speech recognition by introducing caching, a two‑level distributed index, batch processing, and a cascading fault‑tolerance mechanism, demonstrating robust scalability and low communication overhead in Tencent's WeChat ASR service.
This article provides a comprehensive overview of Apache Druid's distributed column‑store architecture, detailing its node types, external dependencies, data flow, and operational mechanisms, and compares Druid's real‑time analytics capabilities with systems such as Impala, Elasticsearch, and Spark.
This article explains the design of a distributed video transcoding platform that leverages AWS Lambda and EC2‑SLAVE to achieve high‑concurrency, low‑latency streaming, detailing architecture, load balancing, health checks, task monitoring, and cost‑saving strategies for scalable cloud‑based video processing.
Tencent Cloud’s overseas video live‑streaming architecture combines tiered carrier procurement across North America, APAC and Europe, decentralized origin servers with dual‑active synchronization, and network optimizations such as DNS‑based scheduling, QUIC and QTCP, cutting stutter rates while tightly controlling costs.
This article explains the evolution and classification of web crawlers, describes the design and components of the Mole distributed crawler—including scheduler, fetcher, processor, rate‑limiting, URL deduplication, and Elasticsearch storage optimization—and outlines common anti‑anti‑crawling strategies.
This article explains Redis Cluster’s distributed architecture, the 16,384 hash slot mechanism, node addition and removal, decentralised request routing, fault‑tolerance through master‑slave replication, slot migration without service interruption, and the internal data structures that enable these features.
This article explains the architecture and Java code for a high‑concurrency flash‑sale system, covering request distribution, pre‑processing, queue selection, transaction handling, and database interaction to efficiently manage limited‑stock purchases under massive traffic.
The article recounts how Xiaomi’s engineering team built, tested, and iteratively improved a high‑concurrency flash‑sale platform—first in PHP with Redis and later in Go—to reliably handle millions of users during the 2014 Mi Fan Festival, detailing architectural choices, performance optimizations, and lessons learned.
This article explains how Weibo’s massive image‑hosting platform uses a cross‑IDC distributed object storage system, optimized upload/download pipelines, and a custom compression library to handle billions of images and extreme traffic spikes during events like the Chinese New Year.
