This article walks through a scenario‑driven capacity assessment for a production‑grade Kafka cluster, covering QPS calculations, storage needs, physical machine count, disk choices, memory, CPU, network bandwidth, deployment steps, and a final resource summary.
This article explains the importance of conducting full‑link load testing in production environments, outlines the evolution and solution architecture, describes key technologies such as traffic coloring, data isolation and risk control, and shares practical implementation steps and customer case studies from Alibaba.
This article explains how the Havok full‑link load testing platform was designed and built to replay real traffic safely, provide capacity‑assessment data, support multiple test types, and offer real‑time monitoring and circuit‑breaker protection for large‑scale online services.
This article compiles core architectural concepts—including ROI, SOLID principles, system splitting, isolation, ACID, CAP/BASE, distributed transactions, and capacity estimation—explaining their definitions, practical examples, trade‑offs, and how they guide architects in making informed technical decisions.
This article explains the SRE role originated by Google, outlines its core responsibilities such as automation, observability, incident response, testing, capacity planning, and SLI/SLO/SLA management, and highlights the skills and cultural practices needed for reliable service operations.
The article presents a case study of how Alibaba Cloud’s network operations team tackled the massive, unpredictable traffic spikes of the 2021 Double 11 shopping festival by identifying transaction‑promotion traffic risks early through AI‑powered analysis, overcoming the limitations of manual rule‑based detection, and achieving precise, automated capacity risk control.
The article reviews the Xi'an “One Code Pass” health‑code platform, covering its award recognition, recent service outages, capacity‑planning calculations, security‑platform procurement, Ministry engineer inspection, and the identified technical bottlenecks such as lack of CDN for static assets and insufficient outbound bandwidth.
This article explains how to design and evaluate system capacity by defining key metrics such as QPS, TPS and concurrency, describing when capacity assessment is needed, and outlining a step‑by‑step methodology—including traffic analysis, peak estimation, stress testing and redundancy planning—to ensure reliable performance under varying load conditions.
This article details the end‑to‑end preparation, execution, reporting, and retrospective of a large‑scale production‑environment load test for the Double 11 shopping festival, covering data preparation, QPS target calculation, multi‑scenario testing, issue analysis, and continuous improvement practices.
This article shares practical experiences and best‑practice guidelines for deploying and operating Prometheus in Kubernetes, covering version selection, inherent limitations, exporter choices, metric design, multi‑cluster scraping, memory and storage planning, GPU monitoring, timezone handling, and alerting considerations.
The article explains the reasons for scaling, outlines hardware and component expansion strategies, introduces the AKF splitting principle for distributed systems, discusses database clustering and distributed ID generation methods such as UUID and Snowflake, and describes elastic scaling challenges and solutions.
This article explains why capacity expansion is necessary, outlines hardware and component scaling strategies, introduces the AKF splitting principle for Redis clusters, discusses challenges of distributed scaling such as data consistency and high concurrency, and reviews database clustering and distributed ID generation methods like UUID and Snowflake.
The article explains why performance optimization has limits and outlines practical scaling methods—including whole‑machine and component upgrades, AKF splitting, database clustering, distributed ID generation (UUID and Snowflake), and elastic scaling—while also discussing the challenges each approach introduces.
This article explains why capacity expansion is necessary, outlines hardware and component scaling strategies, introduces AKF splitting principles, discusses database clustering and distributed ID generation methods such as UUID and Snowflake, and highlights the benefits and challenges of elastic scaling.
The article explains how to assess system capacity by analyzing daily traffic, calculating average and peak QPS, applying the 80/20 rule, conducting stress tests, and adjusting resources, illustrated with a sports event example and a book reservation system case study.
This article explains how to design and evaluate system capacity—covering concepts like QPS, concurrency, peak‑traffic estimation, stress testing, and practical steps to ensure your services can handle both normal loads and sudden traffic spikes without failure.
This article explains why sizing a database cluster is challenging, outlines a systematic capacity‑planning process, examines workload characteristics, query‑operation mapping, consistency trade‑offs, and maintenance considerations, and demonstrates how the open‑source NoSQL database ScyllaDB can be used to model and simplify these decisions.
This article presents a comprehensive, step‑by‑step guide on improving system reliability through early fault detection, scope reduction, frequency reduction, and rapid incident handling, using real‑world practices from Baidu's commercial hosting platform.
This article explains how to estimate and plan system capacity—covering concepts like QPS, concurrency, peak traffic calculation, load testing, and practical steps with examples—to ensure services can handle both regular and surge workloads effectively.
This article explains how to assess and design system capacity by translating a real‑world sports event scenario into concrete metrics such as daily visits, average and peak QPS, concurrency, and instance limits, while outlining practical steps, formulas, and a book‑reservation case study.
This article explains how to design and evaluate system capacity by using a sports event example, defining key metrics such as QPS, TPS and concurrency, applying the 80/20 rule, performing stress tests, and calculating required instances for reliable operation.
This talk shares practical high‑availability strategies learned from years of building Taobao’s massive e‑commerce platform and migrating to Alibaba Cloud, covering traditional IDC stability mechanisms, cache and disaster‑recovery designs, cloud‑native fault‑tolerance, capacity planning, rate‑limiting, graceful degradation, and multi‑region resilience.
Shopee Shake’s architecture separates admin and user sides into three layers—access, application, and resource—and uses horizontal scaling, bucketed Redis coin pools, multi‑level caching, asynchronous message queues, precise capacity formulas, and comprehensive monitoring and chaos‑engineered runbooks to reliably handle over 300,000 QPS during major promotional events.
The article explains why large‑scale services must conduct production full‑link load testing, describes its evolution from ad‑hoc trials to standardized monthly practices, and details the technical and procedural steps—including traffic modeling, JMeter usage, middleware tagging, and responsibility mapping—that ensure reliable capacity planning and risk mitigation.
This document outlines a comprehensive SRE‑driven operational framework for ensuring stable, high‑availability online education services during peak summer and winter periods, detailing pre‑, during‑, and post‑maintenance phases, architectural principles, load testing, monitoring, capacity management, safety hardening, chaos engineering, incident response, and post‑mortem practices.
Alibaba’s eight‑year Double 11 journey illustrates how the company tackled exponential business growth by inventing high‑availability middleware, precise capacity planning, unit‑based deployment, online stress testing, hybrid‑cloud elasticity, and intelligent runtime control to balance throughput, cost, and user experience during the midnight peak.
This article explains how to estimate and design system capacity by analyzing daily traffic, calculating average and peak QPS, applying the 80/20 rule, performing stress tests, and adjusting resources to handle temporary spikes, initial launches, and growing workloads.
This article explains how to design and evaluate system capacity by defining key concepts such as design capacity, TPS, QPS, and concurrency, outlining when capacity assessment is needed, and providing a step‑by‑step methodology with real‑world examples and calculations for accurate performance planning.
The article analyzes WeChat Moments' massive image and video services, detailing its OC/IDC architecture, holiday traffic challenges, software and hardware safeguards, disaster‑recovery mechanisms, retry policies, and a series of flexible strategies—including compression format changes, bitrate reduction, buffer pools, and timeline throttling—to sustain billions of daily accesses.
This article analyzes the large‑scale architecture of WeChat Moments, detailing image and video traffic characteristics, hardware and software safeguards, disaster‑recovery mechanisms, capacity assessment, and a series of flexible strategies such as compression format changes, bitrate reduction, buffer pools, and timeline throttling to handle holiday spikes.
This article explains how to assess and design system capacity by analyzing real‑world scenarios—such as a company sports event—calculating required concurrency, average and peak QPS using the 80/20 rule, performing load tests, and determining instance counts to ensure reliable performance under varying traffic spikes.
This article clarifies key performance metrics such as QPS, TPS, PV, UV, DAU, MAU, concurrent users, and system throughput, explains their differences, relationships, and how they impact capacity planning, while also outlining essential performance testing concepts and evaluation methods for robust system design.
This article explains how to estimate and design system capacity—including traffic, concurrency, QPS, and resource requirements—by illustrating concepts, analytical formulas, peak‑load assessment techniques, performance testing, and a practical case study of a book‑reservation system.
This guide explains Kubernetes capacity planning, showing how to detect idle CPU and memory, identify wasteful namespaces, use open‑source tools like kube‑state‑metrics and cAdvisor, and apply PromQL queries to optimize resource requests and measure the impact of your improvements.
The article explains service governance and SRE practices, detailing goals, components, overload handling, capacity planning, and strategies to maintain continuous 24‑hour service reliability while reducing manual toil.
After migrating to the cloud, enterprises must evaluate resource scale, cost pressure, and staffing before deciding whether to build their own scheduling system, and can choose among ECS, Dedicated Host, or private pool solutions, each with specific advantages, drawbacks, and suitable scenarios.
This article shares practical experiences and best‑practice guidelines for using Prometheus in Kubernetes environments, covering version selection, inherent limitations, common exporters, Grafana dashboards, metric selection principles, multi‑cluster scraping, GPU monitoring, timezone handling, memory and storage planning, and alerting considerations.
This article presents a comprehensive, step‑by‑step framework for guaranteeing system reliability during high‑traffic promotional periods, covering SRE hierarchy, stability criteria, profiling, monitoring, capacity planning, incident response, and post‑event analysis to help teams build resilient services.
To support Gaode Taxi’s soaring order volume, the team expanded from four to eight ECS instances, re‑sharded 256 tables into 4,096, built a custom binlog‑to‑Kafka sync middleware for full‑load and incremental migration, implemented rigorous validation and repair processes, and employed a gray‑rollout with ABC verification, completing the migration without code changes or incidents.
This article shares practical insights from a technical leader on designing robust system architecture, implementing comprehensive capacity planning, establishing reliable operations processes, strengthening security, and cultivating team awareness to achieve long‑term stability for large‑scale internet services.
This article details a systematic full‑link performance testing workflow—including background, timing, scenario design, data preparation, capacity planning, monitoring, issue analysis, and post‑test cleanup—aimed at reliably evaluating and scaling e‑commerce services during major promotional events.
This article details iQIYI's end‑to‑end load‑testing methodology for its payment platform, covering problem identification, core‑link mapping, environment setup, realistic traffic modeling, execution safeguards, results from capacity verification and stress testing, and future plans for a unified testing solution.
This article describes iQIYI's payment team approach to full‑link load testing, covering background challenges, systematic problem exploration, preparation of test environments, traffic modeling, execution safeguards, practical results, and future plans to improve capacity verification and system reliability.
iQIYI’s payment team built a full‑chain load‑testing framework that isolates data, mocks dependencies, constructs realistic multi‑service traffic, and executes protected tests to expose bottlenecks, guide scaling and optimizations, and ultimately ensure reliable payment services during traffic spikes, while planning a unified automation platform.
This article explains how enterprises adopting SRE can design a comprehensive observability platform—covering metrics, logs, and tracing—while also detailing effective incident response, post‑mortem practices, testing, capacity planning, automation tool development, and user‑experience focus to improve overall operational reliability.
This article introduces the Hey load‑testing tool, explains how to install and run it with specific QPS settings, analyzes the resulting metrics and charts, and offers practical tips for identifying bottlenecks such as network bandwidth or CPU usage during capacity planning.
This article explains how to construct a truly high‑availability architecture for modern distributed, cloud‑native services by covering disaster‑recovery principles, capacity planning with realistic load testing, online traffic protection, and systematic fault‑drill practices.
The logistics R&D team spent 62 days preparing for the 618 promotion by analyzing core processes, applying stress tests, implementing fault‑tolerant architectures, planning capacity, and deploying real‑time monitoring tools to ensure system stability and performance under peak traffic.
This article details JD Cloud's comprehensive operational preparation for the 618 shopping festival, covering early resource procurement, hardware fault management, network and CDN scaling, extensive capacity‑testing, disaster‑recovery drills, and cross‑departmental coordination that together ensured stable service during massive traffic spikes.
This article explains what software performance means, how to measure it with metrics such as response time and throughput, how to diagnose problems, visualize time and space contributions, and apply optimization principles, capacity planning, and testing to keep systems performant.
This article reviews the evolution of Alibaba's Aliware message engine, analyzes the low‑latency and high‑availability challenges faced during Double 11, and details the architectural, JVM, memory, rate‑limiting, and multi‑replica solutions that enabled RocketMQ to achieve sub‑millisecond write latency and five‑nine availability.
During Didi’s 2019 Carpool Day promotion, a surge of up to 6.6‑times normal matching traffic forced a redesign of its dispatch engine, introducing near‑time assignment, filtered logic moves, configurable timeouts, extensive stress testing, monitoring, and rapid on‑call procedures that cut downstream pressure by over half.
This article explains why online applications experience crashes during traffic spikes, outlines the complexity of modern cloud‑based service architectures, and shares Alibaba engineers’ practical notes on high‑availability design, capacity planning, full‑link stress testing, monitoring, traffic control, routine inspections, and chaos‑engineering drills using tools such as AHAS, PTS, Sentinel and Advisor.
This article outlines how Ping An’s IT operations team systematically prepares for high‑traffic business events—detailing service assessment, architecture mapping, configuration audits, monitoring design, capacity planning, stress testing, and coordinated incident response—to guarantee reliability and performance under massive concurrent loads.
The article explains how Alibaba introduced systematic capacity planning, four‑stage capacity assessment, various single‑machine stress‑test techniques, and a full‑link stress‑testing platform to reliably handle the massive traffic spikes of the Double‑11 shopping festival, while also describing a flexible traffic‑control framework to prevent overload and avalanche effects.
This article explains how e‑commerce promotions differ from offline sales by offering lower participation thresholds and flexible discount tactics, outlines methods for estimating and handling traffic spikes, and provides detailed strategies for system capacity planning, load testing, monitoring, and incident response to ensure stable large‑scale promotional events.
The article presents a systematic capacity‑estimation methodology that links service traffic to order volume, uses CPU‑Idle as a primary metric, predicts traffic growth and upper‑bound limits, validates predictions with load‑testing, and provides scaling recommendations while noting limitations of the CPU‑Idle baseline.
The article outlines Alibaba Youku’s end‑to‑end resource‑assurance platform for Double‑11 promotions, detailing automated demand collection, business‑to‑technical metric conversion, single‑machine capacity testing, rapid scaling and emergency borrowing, which together cut manual reviews by 80 % and boosted delivery efficiency tenfold.
The article details how Alipay engineers tackled the massive traffic spikes of Double 11 by addressing external payment bottlenecks, implementing recharge‑based balances, building capacity‑planning platforms, adopting logical data‑center (LDC) and CRG zone architectures, deploying elastic scaling, and evolving their OceanBase database and service‑mesh infrastructure to sustain millions of transactions per second.
The article details how 7FRESH's R&D, testing, network operations, and product teams coordinated system capacity expansion, degradation mechanisms, emergency response procedures, extensive performance testing, and 24/7 monitoring to ensure stable and scalable service during the high‑traffic 11.11 shopping event.
Alibaba’s seven‑year journey of full‑chain performance testing for its Double 11 shopping festival reveals a comprehensive, production‑environment‑based workflow—including environment transformation, data preparation, traffic safety, test execution, and intelligent analysis—designed to ensure system stability under massive traffic spikes and guide external customers.
This article explains how to assess, monitor, and tune system I/O performance by defining I/O models, selecting appropriate evaluation tools, tracking key metrics for disk and network I/O, and applying practical optimization strategies for both storage and network bottlenecks.
To guarantee stability for over 100 million daily users, Gaode combines capacity planning, traffic control, disaster recovery, monitoring, and pre‑plan drills with a self‑built full‑link load‑testing platform (TestPG) that replays realistic traffic in production‑like environments, isolates test loads, provides rapid configuration, detailed debugging, automated error capture, and comprehensive reporting, while planning future enhancements such as integrated topology monitoring, advanced pressure models, and confidence evaluation.
This article explains vertical and horizontal database sharding, compares hash-modulo and range-based partitioning, and proposes a hybrid design that balances load, prevents hotspots, and eliminates costly data migrations during scaling.
This article outlines four core problems that backend teams regularly encounter—capacity planning, system stability, monitoring, and development efficiency—and offers practical insights and tool recommendations to help new interns understand and address these challenges.
This article chronicles Alibaba's decade‑long journey of capacity planning for Double 11, detailing five evolutionary phases—from manual estimates to full‑link testing ecosystems—while balancing cost, stability, and efficiency in massive distributed systems.
The JD Finance technical teams—including Wealth R&D, Consumer Finance, Payment, Middle‑Platform, and Crowdfunding—conducted comprehensive system reviews, performance stress tests, capacity expansions, monitoring enhancements, and emergency downgrade plans to ensure stable, high‑throughput service during the 11.11 shopping festival.
The article introduces SGM, a comprehensive service governance and monitoring solution that addresses scaling, dependency complexity, and operational challenges by providing automated topology, real‑time tracing, capacity planning, root‑cause analysis, and extensive monitoring features such as performance metrics, JVM stats, call‑chain visualization, business dashboards, and intelligent alerting.
This article outlines the evolution of Ctrip’s performance testing approaches across three generations, analyzes their limitations, and presents the design, architecture, data construction, request tracing, monitoring, and operational considerations of the fourth-generation full‑link testing platform, including case studies and future outlook.
Tencent’s architecture platform team explains how they monitor, automate, and secure billions of daily operations across storage, CDN, and live services, using multi‑dimensional metrics, real‑time and instant computation, AI‑driven anomaly detection, and a custom control platform for safe changes.
This article offers eight practical operations‑team strategies—inspection, monitoring alerts, capacity planning, network restrictions, risk pre‑plans, data backup, on‑call mechanisms, and staying connected—to ensure system stability and enjoy the Golden Week holiday without incidents.
This article reviews the evolution of Alibaba's Aliware message engine, analyzes the latency and availability challenges faced during Double‑11, and describes the low‑latency optimizations, capacity‑guarantee strategies, and multi‑replica high‑availability architecture implemented in RocketMQ to sustain trillion‑level message flows.
This article explains how to assess I/O capabilities, choose appropriate evaluation tools, monitor key performance metrics, and apply targeted optimization techniques for both disk and network I/O to improve system throughput and latency.
The article explains how Tencent's SNG operations team uses business profiling—including capacity, activity, core‑link, and SET models—to address performance testing across device types, forecast activity‑driven resource needs, identify core versus peripheral services, and plan reliable multi‑region deployments.
This article explains how Tencent's SNG operations team builds and applies business profiling models—including capacity, activity, core‑link, and SET planning—to predict performance, automate scaling, identify critical services, and efficiently distribute workloads across multiple regions.
This comprehensive guide outlines sixteen practical steps—from initial communication with developers to capacity planning and incident response—to help engineers efficiently assume ownership of a new business’s operations while ensuring stability, security, and cost‑effectiveness.
This article explains fundamental performance concepts—defining response time and throughput, using axiomatic methods, analyzing bottlenecks with sequence diagrams and profiling, applying Amdahl’s Law, and guiding capacity planning to build reliable, high‑performance applications.
This article presents a comprehensive case study of Baidu Nuomi's online load‑testing methodology for high‑traffic marketing events, covering capacity estimation, test planning, execution, anti‑attack measures, platform architecture, and lessons learned to ensure system reliability and performance under peak loads.
This article explains JD Finance's operational challenges in a rapidly expanding micro‑service environment and presents a comprehensive approach that combines offline and online load testing, precise capacity calculations, and intelligent root‑cause alert analysis using both rule‑based and machine‑learning techniques.
On December 30, 2017, a sudden wave of users uploading and downloading their 18‑year‑old photos caused QQ Space's album service to experience a four‑times spike in download traffic and a twelve‑times surge in post activity, prompting the operations and development teams to employ capacity monitoring, elastic scaling, flexible architecture, and targeted optimizations to maintain service stability and user experience.
This article explains the purpose of performance (stress) testing, compares four testing approaches, details the chosen proportional‑deployment strategy, and provides comprehensive preparation steps, script guidelines, metric analysis, and practical tips for ensuring system stability during large‑scale traffic spikes.
Alibaba’s full‑link stress testing, which simulates real‑world traffic across the entire e‑commerce platform, enabled the 2017 Double 11 event to handle peak loads of 325,000 transactions per second, demonstrating how production‑level, data‑isolated load testing ensures stability and capacity planning for massive online sales.
Suning’s 2017 O2O shopping festival achieved a “zero‑incident” goal by integrating real‑time browser performance monitoring, WEEX‑based WAP acceleration, comprehensive app data collection with cloud‑based analytics, precise DNS and HTTP2 optimizations, and a multi‑layer network and service monitoring system that enabled rapid fault detection and capacity planning.
This article details Ele.me's rapid expansion challenges and shares a three‑stage technical operations journey—fine‑grained division, stability maintenance, and efficiency gains—highlighting real incidents, monitoring upgrades, capacity testing, and practical insights for reliable large‑scale delivery platforms.
This article explains why service monitoring is essential for large‑scale microservice environments, outlines design principles, core monitoring components, dependency mapping, call‑chain analysis, capacity planning, root‑cause analysis, and presents a practical technical architecture for implementing robust monitoring solutions.
This article explains how Tencent's social platform operations team prepares for massive holiday traffic spikes by following a four‑stage process—business preparation, capacity evaluation, resource provisioning, and scaling with stress testing—while detailing team structures, operational standards, and the supporting tool ecosystem that enable reliable, high‑availability services.
Alibaba’s eight‑year evolution of Double 11 showcases how limited cost can deliver maximal user experience and massive throughput by transitioning from a centralized 3.0 distributed architecture to multi‑active zones, employing capacity planning, full‑link stress testing, fine‑grained dependency governance, and dynamic traffic scheduling to ensure high availability.
The article shares practical operations lessons—from periodic permission audits and thorough documentation to capacity monitoring, log rotation, and automation—illustrating how systematic practices and tooling can standardize and streamline IT infrastructure management.
This article explores how Tencent’s operations team redefines its value by applying fine‑grained technical management to mobile internet challenges, capacity planning, bandwidth optimization, and data‑driven product decisions, ultimately delivering huge cost savings and turning operations into a core competitive advantage.
This article examines JD.com’s 618 shopping festival performance, the deployment of unmanned delivery robots, and the design and architecture of the ForceBot full‑link load‑testing system that enabled precise capacity planning and bottleneck detection for massive e‑commerce traffic.
This talk shares practical high‑availability designs learned from Alibaba's Taobao platform and Alibaba Cloud, covering traditional IDC stability mechanisms, modern cloud‑native fault‑tolerance, caching tricks, performance tuning, limit‑and‑degrade tactics, disaster‑recovery planning, and multi‑region deployment for handling billions of requests during peak events.
This article explains Alibaba's four‑step capacity‑planning methodology, the various single‑machine load‑testing techniques, the design of a full‑link load‑testing platform for Double‑11, and the dynamic flow‑control framework that together ensure system stability during extreme traffic spikes.
This guide walks through the evolution of a small .NET‑based e‑commerce system, covering its initial LAMP‑style setup, detailed backend architecture, logging and monitoring solutions, master‑slave database design, shared‑storage image server, mobile M‑site construction, capacity estimation methods, and caching strategies.
This article presents a comprehensive methodology for internet performance engineering, covering non‑functional quality goals, detailed metrics for application servers, databases, caches and message queues, a practical technical review outline, and a real‑world capacity‑planning case study with both maximal and minimal resource solutions.
This article presents a comprehensive methodology for reviewing internet‑scale system architectures, focusing on non‑functional quality attributes such as performance, availability, scalability, security, and maintainability, and provides detailed guidelines, metrics tables, and a classic case study for capacity and performance planning.
This article chronicles Alibaba's evolution of the full‑link pressure testing platform—from its 2013 inception tackling massive Double 11 traffic, through data construction, isolation, traffic generation, and platform upgrades—to a mature, automated, cloud‑native solution that safeguards large‑scale e‑commerce stability.
This article outlines a comprehensive PDCA‑based methodology for e‑commerce platforms to proactively prevent issues, quickly detect anomalies, and execute rapid decisions during large‑scale promotions, covering system goal definition, performance evaluation, capacity planning, SLA management, and team/process maturity.
The article explains how LinkedIn’s Redliner system automatically measures service capacity and performs low‑impact, production‑traffic stress tests to identify bottlenecks, guide resource allocation, and support proactive capacity planning and performance regression testing.
The talk details how Alibaba tackled the massive technical challenges of Double 11 over eight years by evolving a highly available, scalable architecture through capacity planning, distributed middleware, hybrid‑cloud deployment, online stress testing, and fine‑grained traffic control to balance cost, performance, and user experience.
This article details Alibaba's eight‑year journey of scaling its Double 11 shopping festival by evolving high‑availability middleware, capacity planning, unit‑based architecture, hybrid cloud and automated traffic control to achieve reliable, cost‑effective performance at massive peak loads.
