This article explains how enterprises can leverage cloud‑native architectures and FinOps practices to gain financial accountability, visualize multi‑dimensional cost data, optimize resource usage, and implement systematic cost governance across Kubernetes environments, covering cost insight, optimization, and operational stages with practical recommendations and example algorithms.
This article details the end‑to‑end design of Vivo’s custom log‑collection agent, covering file discovery with inotify, unique file identification using inode and content hash, real‑time reading via RandomAccessFile, checkpointing, Kafka integration, offline HDFS ingestion, resource throttling, and platform‑wide management, while comparing it with open‑source alternatives.
This article explains how Go's defer statements evaluate arguments, interact with panic, affect return values, and behave with os.Exit, providing clear code examples and conclusions to help developers avoid common mistakes.
This article presents a detailed case study of a leading AI‑driven quant investment firm that leveraged Alibaba Cloud ACK's FinOps suite to tackle planning, allocation, management, and optimization challenges in Kubernetes‑based container workloads, achieving up to 25% cost reduction and significantly higher resource utilization.
This article presents a systematic approach for technology leaders to reduce and control cloud‑related expenses through cost‑optimization, detailed analysis, team organization, project planning, communication, and long‑term fine‑grained operational practices.
Koordinator extends its open‑source container scheduler to enable seamless co‑location of Kubernetes Pods and Hadoop YARN tasks, allowing over‑provisioned batch resources to be shared without modifying YARN, and has delivered up to 10 % CPU utilization gains and sub‑1 % eviction rates in Xiaohongshu’s production clusters.
The article explains how to use throttling together with auto‑scaling to control resource consumption of cloud applications, prevent tenant overload, handle traffic bursts, and ensure service‑level agreements while optimizing costs.
This article explains Kubernetes Limits and Requests, their differences, practical deployment examples, CPU and memory characteristics, namespace ResourceQuota and LimitRange policies, and provides best‑practice guidelines to optimize resource allocation, avoid OOM kills, and prevent CPU throttling.
The article explains how Koordinator and containerd integrate the Node Resource Interface (NRI) to improve container QoS, avoid noisy‑neighbor issues, and provide a flexible, standardized resource‑management model for cloud‑native workloads, as presented at KubeCon China 2023.
Uber reduced cluster-wide CPU allocation by 11% and stabilized P99 latency by replacing CPU quota throttling with cpusets, a CPU‑pinning technique that isolates containers on specific cores, eliminating throttling while only slightly increasing P50 latency.
The article broadens software architecture beyond structural design to encompass people, processes, and business value, urging architects to balance realistic, flexible goals, limited resources, commercial‑focused behaviors, and industry trends so they can deliver cost‑effective, scalable solutions that align technology with organizational objectives.
This article demonstrates practical ways to leverage Java's ShutdownHook API for task statistics, connection cleanup, handling JVM abnormal exits, and persisting state during performance testing, providing code examples, pitfalls of daemon threads, and strategies for reliable resource management in backend applications.
This article presents a plug‑in‑based, WebWorker‑enabled resource manager for web games and live‑stream activities, detailing its background, research of existing solutions, overall design, core modules, external APIs, and practical usage examples such as preloading, modular bundles, and image conversion.
This article explains Kubernetes resource management, covering compute and non‑compute resources, key mechanisms such as quotas and autoscalers, monitoring tools, and optimization strategies to improve cluster efficiency, scalability, and cost effectiveness.
The article explains why enterprises adopt multi‑cluster Kubernetes architectures, reviews community solutions such as Karmada, Clusternet and OCM, and details vivo’s hybrid strategy that combines a unified UI for independent clusters with Karmada‑based federation for resource distribution, elastic scaling, cross‑cluster scheduling, and gray‑release migration.
The MagicScaler framework, introduced by Alibaba Cloud’s big‑data engineering team and collaborators, combines a multi‑scale attention Gaussian process predictor with an uncertainty‑aware elastic scaling decision engine, delivering significantly higher quality‑of‑service and lower operational costs than traditional autoscaling methods, as demonstrated on real MaxCompute workloads.
ByteDance’s multi‑cloud evolution, driven by rapid business growth, cost control, and compliance needs, showcases a distributed cloud‑native platform built on open‑source orchestration, unified resource management, and advanced data‑lake solutions, while addressing operational complexity, interoperability, and emerging AI‑driven challenges.
The article details the end‑to‑end technical implementation of Bilibili’s “Da Li Chu Qi Ji” mini‑game, describing a GSAP‑driven 4‑second countdown, a PixiJS‑based core gameplay with custom resource loading and event handling, and the evaluation and integration of PAG, SVGA, and MP4 animation formats for reward effects.
This announcement introduces Huawei Cloud’s DTT live session on August 29, where expert Lv Pengbo will explain GaussDB(DWS) resource control principles and demonstrate practical techniques—such as CPU usage analysis, memory tuning, and queue issue resolution—to help developers efficiently manage data‑warehouse resources and boost performance.
This article details how ZEEK automotive tackled rapid growth challenges by redesigning its cloud‑native infrastructure, adopting Alibaba Cloud ACK FinOps and ACK One for multi‑cluster management, and implementing cost‑visibility, intelligent allocation, and configuration checks that yielded significant resource savings and operational stability.
Katalyst v0.3.0 introduces major upgrades including enhanced KCNR API bandwidth isolation, a more extensible task and async execution framework, advanced mixed‑deployment controls, load‑aware resource prediction, and concurrent unit testing, all aimed at improving cloud‑native resource management efficiency.
By integrating Kubernetes with DevOps practices, teams can automate deployment pipelines, achieve dynamic resource allocation, centralize monitoring with tools like Prometheus and Grafana, and treat infrastructure as code, resulting in faster, higher-quality software delivery and improved collaboration between development and operations.
The article explains that an operating system acts as a resource allocator, isolates processes to maintain system order, and abstracts hardware complexities, using examples of CPU scheduling, memory management, and networking to illustrate why OSes are essential for modern computing.
This article explains Python's context manager protocol, detailing the roles of __enter__() and __exit__() methods, their typical use cases, and provides ten concrete code examples ranging from file handling and database connections to custom managers for networking, temporary files, and exception handling.
The article explains why testing teams need systematic public resource management, outlines the challenges of resource flow, asset control, and security, and details the evolution from a simple approval pipeline (v1.0) to a more autonomous, domain‑aware system (v2.0) with practical solutions.
This article details the current challenges, governance motivations, architectural design, and technical optimizations—including Flink SQL tuning, Kafka batch improvements, partitioned stream tables, containerization, and automated governance—implemented to enhance the efficiency, stability, and cost-effectiveness of NetEase Cloud Music's real-time data warehouse platform.
This article provides a comprehensive overview of Linux cgroups, explaining their purpose, architecture, versions, subsystems, and practical usage with systemd, including installation, configuration files, command‑line tools, and methods to monitor and limit CPU, memory, and I/O resources for containers and services.
This article explains Kubernetes QoS classes—Guaranteed, Burstable, and BestEffort—detailing their resource request and limit requirements, how to configure them with YAML examples, the eviction priority order, and practical best‑practice strategies for classifying and scheduling workloads across cluster nodes.
This article explains ByteDance’s cloud‑native mixed‑workload strategy, describing how the open‑source Katalyst system leverages Kubernetes and Yarn to dynamically share idle resources between online and offline services, improve utilization from 23% to 60%, and reduce operational costs.
This article explains how Kubernetes uses CPU requests and limits to schedule pods, allocate CPU proportionally, calculate minimal request units, and provides practical guidelines for setting appropriate request and limit values based on workload characteristics and monitoring data.
This article introduces FinOps, explains its definition and cultural significance, illustrates the concept with a personal household billing story, and outlines five concrete steps—cost allocation, target setting, budget control, operational management, and intelligent automation—to achieve cloud cost optimization.
This article details Hupu's massive sports‑traffic environment, the design and implementation of the Flex auto‑scaling platform, its architecture, core functions such as resource statistics, node and pod scaling, scenario scheduling, and the performance optimizations that enable rapid, cost‑effective scaling across multi‑cloud Kubernetes clusters.
This article examines why front‑end teams often become mere resources, outlines the emotional and operational challenges of resource‑centric work, and presents practical strategies—including goal setting, role clarification, and risk management—to transform front‑end into a value‑adding leader, illustrated by an experience‑driven pricing revamp case study.
Cougar is a cloud‑native, multi‑objective optimization framework that unifies metadata and monitoring ingestion to improve resource efficiency and performance for large‑scale AI and big‑data jobs, demonstrating over 50% CPU‑memory savings and stable latency in production experiments.
This article examines why Baidu's super‑app needed a drastic reduction in its iOS package size, outlines the metrics that link bundle size to download conversion, compares the app‑size footprints of major domestic and overseas apps, and details the multi‑layered technical solutions—resource trimming, architecture safeguards, compiler tweaks, image compression, and code slimming—that together saved over 50 MB while preserving functionality.
This article explains the object pool design pattern, its working mechanism, step‑by‑step Java implementation using Apache Commons Pool, advantages, drawbacks, and real‑world scenarios such as web server connection pooling and game object reuse, providing complete code examples.
This article details how Zhuanzhuan transformed its testing infrastructure from a handful of monolithic servers to a Docker‑driven, tag‑routed dynamic and stable environment, addressing resource shortages, waste, and stability issues while achieving significant reductions in deployment time, resource consumption, and user‑reported problems.
The article explains Linux cgroup architecture and initialization in kernel 5.10, covering its hierarchical composition, key data structures like css_set, the two‑phase boot‑time setup, creation of cgroups, and task assignment mechanisms for both cgroup v1 and v2.
This article explains how Apache YARN and Google Kubernetes address the three core problems of resource utilization, task responsiveness, and flexible scheduling in distributed environments, detailing their architectures, scheduling models, and practical implications for modern big‑data and cloud workloads.
This article explains the evolution from monolithic to microservice architectures, introduces Kubernetes as the core cloud‑native platform, and details its components, design principles, and resource management strategies for compute, networking, and storage within a cluster.
This article explains Linux cgroups, why they are essential for containerization, their architecture, subsystems, and practical commands for viewing, configuring, and limiting CPU, memory, and I/O resources using systemd and the cgroup filesystem.
This article explains how to use the Spine 2D skeletal animation framework to implement a flexible, runtime face‑customization and outfit‑changing feature on mobile platforms, covering basic concepts, code examples, resource handling, memory optimizations, and platform‑specific integration challenges.
This article analyzes common Jenkins high‑availability challenges, reviews existing industry solutions, and presents Vivo's own Jenkins Scheduler architecture—including API‑gateway, event center, scheduling algorithms, flow‑control, and callback mechanisms—demonstrating its production deployment and future container‑based evolution.
The paper presents Vivo’s Jenkins Scheduler, a master‑centric, high‑availability solution that replaces single‑master Jenkins by integrating an API gateway, event‑driven failure detection, label‑based multi‑dimensional scheduling, Redis/MySQL‑backed flow control, and callback monitoring, thereby balancing resources, enabling rapid failover, persisting queues, and improving build reliability, with plans to containerize Jenkins for Kubernetes workflows.
This article presents a comprehensive guide to software performance optimization, covering low‑level resource exhaustion, horizontal scaling, sharding, lock‑free techniques, and system‑wide strategies, while offering practical examples and references for developers seeking to improve efficiency and scalability.
This article explains why the default Apache Commons‑Pool2 APIs cannot limit idle objects per key, and demonstrates three practical techniques—scheduled cleanup, factory modification, and priority‑based eviction—to enforce per‑key idle limits in a GenericKeyedObjectPool implementation.
This guide explains the advantages of cgroup v2 over v1, outlines the Linux distribution and kernel requirements, and provides detailed, command‑line instructions for enabling cgroup v2 on Debian, configuring kubelet and containerd, and updating monitoring tools for Kubernetes clusters.
This article explains how distributed storage, computing, and resource‑management technologies have evolved—from early Google File System research to Hadoop, Spark, and Kubernetes—enabling enterprises to tackle the 4 Vs of big data while reducing cost, improving performance, and supporting real‑time analytics.
This article demonstrates how to deploy OceanBase 4.X in a minimal single‑node configuration, explains the key resource parameters, provides the necessary YAML configuration and command‑line steps to start the server, create a MySQL‑compatible tenant, and verify resource usage and basic database operations.
This article examines Huolala's comprehensive cloud cost governance strategy, detailing infrastructure distribution, cost‑trend analysis, optimization tactics such as storage, networking, logging, spot instances, containerization, reserved instances, and the role of the Lcloud platform in achieving measurable savings and operational efficiency.
CNStack 2.0 is a cloud‑native PaaS platform built on Kubernetes that unifies resource and workload management, offering agile, open, and secure multi‑cluster capabilities through modular cloud services, a unified API gateway, and integration with open‑source projects such as Sealer, Emissary‑Ingress, cert‑manager, Velero, and OCM.
The article presents an in‑depth overview of Tencent's self‑developed Unified Scheduling Platform, detailing its system architecture, design challenges, performance optimizations, resource‑fair scheduling mechanisms, operational metrics, future roadmap, and a Q&A session that together illustrate how the platform enables massive offline data processing at scale.
This article presents the background, requirements, architecture, key modules, and practical impact of OPPO's non‑intrusive big‑data diagnostic platform—named Compass—designed to quickly locate issues, provide optimization suggestions, and achieve cost‑saving and efficiency gains for large‑scale Spark and Hadoop workloads.
The article explains how wide tables are maintained without excessive resource consumption, outlines criteria for deciding which metrics belong in a wide table, describes their granularity and timeliness considerations, and clarifies that they do not automatically expand when tracking points change.
This article presents the design principles, core techniques, and practical solutions of Vivo's self‑developed Bees log collection agent, covering file discovery, unique identification, real‑time and offline ingestion, checkpointing, resource control, platform management, and a comparison with open‑source alternatives.
Bilibili’s S12 technical guarantee team coordinated dozens of engineering groups, performed resource estimation, built a shared resource pool, applied chaos engineering, high‑availability architecture, and systematic performance testing to ensure the League of Legends World Championship livestream remained stable and responsive under peak traffic.
This article presents a comprehensive overview of Baidu’s ARIES storage platform, detailing its design philosophy, architecture, key concepts, and engineering challenges, and explains how it underpins Baidu Netdisk’s massive data‑plane storage with high availability, cost‑performance trade‑offs, and robust monitoring.
The article details a Tencent‑led, end‑to‑end cost‑optimization project for large‑scale Kubernetes/TKE clusters that collected extensive workload metrics, applied VPA/HPA enhancements, custom scheduling and node‑downscaling via the open‑source Crane platform, ultimately delivering up to 70% CPU and 50% memory savings with zero‑fault deployments.
Linux cgroups, a kernel mechanism for grouping and controlling processes, enable fine-grained resource allocation, monitoring, and isolation, forming the foundation of container technologies like Docker and Kubernetes; this guide explains their concepts, hierarchies, subsystems, versions, configuration, and practical usage on CentOS.
DeepScaling is a deep‑learning‑driven autoscaling framework that predicts workload, estimates CPU usage, and makes reinforcement‑learning‑based scaling decisions to keep microservice CPU utilization at a target level, thereby reducing resource waste while meeting SLOs in large‑scale cloud environments.
The article shows how a unified abstraction called statefulMap, together with a resource‑aware mapWithResource primitive, lets developers implement a wide range of complex reactive‑stream operators—such as buffering, indexing, deduplication, and safe DB access—in a concise, composable, thread‑safe manner, dramatically reducing boilerplate code.
This article provides a comprehensive guide for technical leaders on reducing and managing technology costs through a two‑stage approach of cost optimization and fine‑grained operations, covering team formation, current‑state analysis, discount and storage tactics, project planning, communication, and long‑term process and system support.
Kubernetes 1.25 officially ships cgroup v2 support, enabling the kubelet to leverage the latest Linux resource‑management features, and the article explains what cgroups are, the advantages of v2, how to enable it, migration considerations, and where to learn more.
This article explains how Meituan built a unified, highly available pipeline engine that supports nearly 100,000 daily executions across dozens of services with 99.99% success, detailing the challenges faced, the architectural decisions made, and the future roadmap for further scalability and cloud‑native improvements.
This article details Meituan's three‑year journey building a self‑developed pipeline engine that now handles nearly 100,000 daily executions with over 99.99% reliability, covering background, challenges, architectural decisions, core scheduling and resource‑pool designs, component layering, and future cloud‑native plans.
This article details how Xingdou Cloud leverages cloud‑native techniques such as massive node overselling, custom HPA (SophonHPA), priority‑based QoS, intelligent cleanup, and quota management to achieve dramatic cost reduction and efficiency gains across its multi‑cloud platform.
This article theoretically outlines seven major Java performance optimization strategies—including reuse, computation, result set, resource conflict, algorithm, high‑efficiency implementation, and JVM tuning—explaining their principles, typical techniques, and how they collectively improve resource utilization and application speed.
This article provides a comprehensive overview of the big data ecosystem, detailing nine core technology categories—from data collection and storage to computation, analysis, scheduling, and underlying infrastructure—along with tool comparisons, selection guidelines to help readers quickly build a complete big data knowledge system.
This article explains how Docker containers rely on Linux cgroups and namespaces for resource isolation, details CPU share and quota scheduling, and shows practical commands to inspect cgroup assignments, helping developers optimize container performance on Kubernetes.
This article details Meituan's three‑year journey building a self‑developed, distributed pipeline engine that now handles nearly 100,000 daily executions across dozens of services with over 99.99% reliability, covering the challenges faced, architectural decisions, scheduling and resource‑pool designs, and future cloud‑native plans.
This article introduces Tencent Music's cloud‑native, one‑stop machine learning platform, detailing its engineering workflow, distributed acceleration, inference closed‑loop, edge computing capabilities, and future plans, while highlighting challenges of traditional ML pipelines and the platform's solutions for resource orchestration, storage, scheduling, and GPU utilization.
This article analyzes the architecture and performance bottlenecks of a mobile live‑streaming classroom, defines measurable APM metrics, identifies root causes such as CPU, memory, GPU contention and signaling issues, and presents concrete optimization techniques—including independent processes, containerization, dedicated signaling channels, rendering and thread improvements—that dramatically reduce memory, CPU and frame‑rate problems.
The article outlines defensive programming concepts, emphasizing input validation, error handling, resource management, isolation techniques, and design considerations such as thread safety, cache strategies, and interface versioning to build robust and resilient software systems.
Shilong Fei from Xiaomi Data Platform presents an in‑depth exploration of elastic scheduling for Hadoop YARN, covering background, design of resource pools, auto‑scaling architecture, challenges such as job stability and user transparency, achieved cost reductions, and future plans for further optimization.
This article explains the design and operation of an external resource manager that improves overseas e‑commerce service stability by employing CDN, local memory, and disk caches, detailing access methods, cache mechanisms, CDN configurations, and practical use cases such as Google Fonts and translation integration.
This article explains how Alibaba Cloud's open‑source Koordinator system tackles mixed‑workload challenges by introducing priority and QoS models, resource overcommit, load‑aware scheduling, fine‑grained CPU orchestration, and upcoming features such as GPU scheduling and resource recommendation, all illustrated with architecture diagrams and code examples.
This article details the backend architecture, dialogue engine design, resource estimation, and performance optimization techniques of an AI interview robot used in 58.com’s recruitment platform, illustrating how multi‑branch dialogue flows, RTP communication, session management, and monitoring enable scalable, stable, and efficient online interview services.
The article details Bilibili's evolution of its offline computing platform from Hadoop‑based Hive to Spark, describing migration tools, SQL conversion, result and resource comparison, shuffle stability, small‑file handling, runtime filters, data skipping, ZSTD support, Hive Metastore federation, traffic control, and future optimization directions.
This article details how Alibaba’s DX (DinamicX) framework attains near‑native performance through server‑side XML compilation, lightweight virtual trees, pipeline asynchronous execution, async drawing, and off‑screen resource control, and presents measurable gains such as 40% reduction in main‑thread work and 65% CPU savings on media‑heavy pages.
This talk presents Xiaomi's design and deployment of an elastic scheduling system for Hadoop YARN, covering background analysis, resource‑pool strategy, auto‑scaling architecture, stability challenges, label‑based resource isolation, Spark shuffle handling, cost‑saving results and future plans.
This article explains the purpose and mechanics of Python's with statement, showing how it leverages context managers to replace verbose try‑finally blocks, and demonstrates both class‑based and decorator‑based implementations with clear code examples.
This article explains how Alibaba Cloud Container Service for Kubernetes (ACK) uses container‑level resource profiling with half‑life sliding windows and quantile algorithms to automatically recommend accurate CPU and memory requests, improving cluster utilization while maintaining application stability.
This article details Car Home's AutoStream platform evolution from Storm to Flink‑based versions, covering real‑time application scenarios, strict budget‑controlled resource management, automatic scaling, lake‑house architecture with Iceberg, PyFlink integration, and future plans for resource optimisation and batch‑stream unification.
This article presents a hands‑on SRE framework covering the full product lifecycle—code development, resource planning, deployment, operational reliability, and decommissioning—derived from real‑world practices at Xiaomi and Sina to help teams manage massive internet services efficiently and cost‑effectively.
While lazy loading can reduce initial page load time and save resources, overusing it can slow scrolling, cause layout shifts, hinder SEO, and ultimately degrade overall web performance, so developers need clear guidelines on when and how to apply it effectively.
This article explains common causes of resource waste in Kubernetes clusters, such as over‑provisioned requests and fluctuating workloads, and provides practical methods—including proper request/limit settings, ResourceQuota and LimitRange policies, node affinity, taints and tolerations, and HPA—to improve overall resource utilization and cluster stability.
This guide explains how to configure CPU and memory requests and limits for Kubernetes pods, implement QoS classes, use LimitRange and ResourceQuota, and monitor resource usage with Prometheus queries and Grafana dashboards to ensure stable cluster operations.
This article examines how LinkedIn tackled severe scheduling slowdowns when its Hadoop YARN cluster grew to nearly 10,000 nodes, analyzes the root causes of resource‑manager bottlenecks, and describes the fairness‑redefinition and scheduling‑logic patches that restored throughput and scalability.
This article explains the concept of pooling technology, its advantages, and its practical applications in Java—including thread pools, memory pools, database connection pools, and HttpClient connection pools—while highlighting how these techniques improve performance, resource utilization, and system stability.
This article delves into Android resource fundamentals, classification, referencing, compilation, shrinking, and practical governance techniques, illustrating how to detect and resolve conflicts, unused assets, missing class references, and hard‑coded texts to improve build size and runtime stability.
Learn how Linux namespaces provide resource isolation and how cgroups enforce resource limits, with step‑by‑step commands, code examples in Go, and practical demonstrations of CPU sharing using unshare, cgcreate, cgset, and cgexec to control process behavior.
iQIYI’s cpu‑share‑syncer daemon runs on every node, reads a pod’s iqiyi.com/cpu‑share annotation, updates the pod’s cpu.shares after disabling the Kubernetes CPU CFS quota, and lets high‑priority video‑production pods pre‑empt CPU from lower‑priority pods, significantly speeding task execution.
Move semantics in C++11 let objects transfer ownership of resources via rvalue references and std::move, eliminating costly copies, improving performance, requiring explicit move constructors and assignment operators—preferably marked noexcept—to keep source objects valid, enable container optimizations, and work with NRVO when copying would otherwise occur.
This article explains Alibaba Cloud Container Service for Kubernetes's differential SLO approach, detailing the reclaimed‑resource model, CPU burst and topology‑aware scheduling, kernel group identity, memory watermark tiering, and real‑world case studies that demonstrate significant improvements in cluster efficiency and latency‑sensitive workload performance.
AutoKH is a cloud‑native mixed‑workload framework that integrates Kubernetes and Hadoop to dynamically schedule online and offline tasks, improve CPU and memory utilization, enforce priority classes, and ensure service stability through operators, CronHPA, and resource‑control components.
This article compiles a comprehensive list of common production problems encountered with Apache Flink, covering cluster sizing, checkpoint failures, backpressure analysis, resource allocation, deployment errors, UDF definitions, data skew, Kafka configurations, and provides detailed troubleshooting steps and best‑practice recommendations.
The article details Tubi's Multimedia Processing Platform (TMPP), describing its architecture, processing stages, resource management, and distributed task scheduling for large‑scale video transcoding and delivery across multiple devices.
This article explains the problem of resource ID mismatches when Android plugins use host resources, reviews existing resource‑fixation solutions, and presents a novel “no‑resource‑fixation” method that dynamically maps host IDs at compile‑time and runtime to keep plugins stable across host updates.