When a data platform server suddenly shows CPU usage above 90%, this guide walks you through using Linux tools and a custom script to identify the offending Java process, trace the problematic thread, pinpoint the exact code line, and apply a fix that reduces load dramatically.
This article explains MySQL index covering and how avoiding table lookups (back‑to‑table) can reduce retrieval steps, improve query performance, and provides practical examples, execution‑plan analysis, and optimization tips for large tables in production.
By profiling the OCI runtime with hyperfine and applying a series of kernel patches, seccomp hash optimizations, and BPF compilation caching, the author reduced container start‑up time from roughly 160 ms in 2017 to just over 5 ms today—a near 30‑fold speedup.
Learn how to effectively use React Context for state management, avoid props drilling, and eliminate redundant re‑renders by implementing a custom publish‑subscribe store, selector hooks, and performance optimizations similar to libraries like zustand.
This article details the technical design, challenges, and results of Taobao’s “Home‑Page as Venue” initiative, explaining how a new A/B‑page architecture replaced crowded promotional cards with a dynamic cat‑head entry, improved information‑flow efficiency, reduced development cost, and delivered measurable gains in traffic, conversion, and revenue during major sales events.
This interview explores the evolution of QQ Browser from its 2009 launch, the technical challenges of building a mobile web rendering engine, the role of Tencent Browser Service (TBS) in document and media handling, performance and security optimizations, market positioning, and future product strategies.
The article presents a JSON‑driven solution for rendering static text templates in video editing, detailing asset integration, cross‑platform layout using CoreText and FreeType, asynchronous drawing, memory and performance optimizations, and guidelines for consistent rich‑text rendering across iOS, Android, and web.
At the GTC conference, the talk presents Alibaba Cloud’s heterogeneous computing platform and introduces the Open Deep Learning API (ODLA), then details how CUTLASS‑based operator fusion dramatically accelerates attention and MLP layers in large‑scale recommendation models, achieving multi‑fold performance gains in production.
This extensive tutorial walks through Nginx concepts, environment setup, reverse‑proxy load balancing, static‑dynamic separation, resource compression, buffering, caching, IP whitelist/blacklist, CORS handling, anti‑hotlinking, large‑file configuration, SSL certificate setup, high‑availability with Keepalived, and key performance optimizations.
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 explains how to detect and resolve performance bottlenecks in Flutter apps by using compile modes, profiling tools, layout tree insights, widget build optimizations, const constructors, list view strategies, raster thread tuning, and memory‑saving practices such as oversized‑image detection and layer separation.
The article describes NetEase Cloud Music’s custom client‑side APM system that combats sliding stutter, heating, UI freezes and crashes by employing binary‑tree stack aggregation to halve storage, window‑based CPU analysis, run‑loop jank detection, ping‑based ANR monitoring, and malloc‑logger memory tracking with automated dump thresholds.
This article explains how to assess MySQL table size, identify performance issues caused by massive data volumes, and presents three practical solutions—table partitioning, database sharding, and hot‑cold data archiving—along with detailed SQL examples and considerations for implementation.
This article analyzes why MyBatis‑Plus's default saveBatch method does not achieve true batch insertion, explains how to enable JDBC batch mode with rewriteBatchedStatements, and demonstrates creating a custom SQL injector (InsertBatchSomeColumn) to perform real batch inserts, including performance test results showing significant speed improvements.
This article analyzes Baidu’s ERNIE‑ViLG 2.0 and PaddlePaddle‑optimized Stable Diffusion models, presenting benchmark comparisons, hardware‑specific speed and memory gains, and the underlying inference optimizations that enable low‑cost, high‑throughput AI‑generated image creation.
The article analyzes the growing performance challenges of data warehouses, evaluates traditional solutions such as clustering, pre‑computation and optimization engines, and presents esProc SPL as a non‑SQL, low‑complexity alternative that delivers orders‑of‑magnitude speedups on modest hardware.
The article details Huolala’s comprehensive approach to managing diverse databases across multiple clouds, covering systematic stability engineering, automated SQL governance, Redis and Kafka reliability strategies, cost‑optimization tactics, and the evolving roles of modern DBAs in a complex, high‑scale environment.
This article details Alibaba Mama’s dynamic compute system—its architecture, offline and online tidal‑compute mechanisms, city‑level mutual backup, RT control, large‑scale promotion handling, metric integration, and recent infrastructure upgrades—showcasing concrete performance gains and future challenges in green, intelligent ad‑engine resource management.
By isolating CPU preprocessing and post‑processing from GPU inference into separate processes and applying TensorRT’s FP16/INT8 optimizations, the custom Python framework boosts Python vision inference services from roughly 4.5 to 27.4 QPS—a 5‑10× speedup—while reducing GPU utilization and cost.
This article outlines Google's most impactful 2023 recommendations for improving Core Web Vitals—Largest Contentful Paint, Cumulative Layout Shift, First Input Delay, and Interaction to Next Paint—providing practical, high‑priority techniques such as ensuring LCP resources are discoverable, prioritized, and optimizing TTFB, CLS, and JavaScript performance.
This article explains why large MySQL tables cause slow queries, analyzes table size, disk usage, and instance capacity, and presents three practical solutions—table partitioning, database sharding (horizontal/vertical), and hot‑cold data archiving—helping developers choose the appropriate strategy for their workloads.
This article details the architectural evolution of JD's VOP message warehouse, describing the challenges of massive data volumes, the transition from V1.0 to V3.0+ through database sharding, MongoDB adoption, traffic governance, stability improvements, and cost reduction strategies, while presenting performance metrics and future outlook.
This article details how the Oasis Engine team tackled the technical and artistic challenges of building a large‑scale, multiplayer, third‑person virtual park for the Alipay Fortune event, covering architecture, scene splitting, asset loading, custom materials, digital human integration, networking, behavior management, and extensive performance optimizations.
This article summarizes Wang Chuanting’s DTCC2022 talk on Huawei Cloud GaussDB(DWS) 3.0, detailing its cloud‑native architecture, layered elasticity, lake‑warehouse integration, performance acceleration techniques, and how it seamlessly couples data‑processing pipelines with AI workloads for modern, real‑time analytics.
This article explains how to assess MySQL table size, why massive tables degrade performance due to B+‑tree depth, and presents three practical solutions—table partitioning, database sharding, and hot‑cold data archiving—along with guidance on selecting the appropriate approach.
This comprehensive guide walks you through Nginx fundamentals, environment setup, reverse‑proxy load balancing, static‑dynamic separation, resource compression, buffering, caching, IP whitelist/blacklist, cross‑origin handling, anti‑hotlinking, large‑file transfer, SSL configuration, high‑availability with Keepalived, and performance‑tuning techniques for production‑grade deployments.
This article describes how Tencent's Alluxio-based Data Orchestration Platform (DOP) was applied to financial analytics, detailing the business background, challenges of large‑scale OLAP workloads, the Alluxio architecture and usage modes, performance results, and the series of optimizations and tuning performed to achieve significant speedups.
This article outlines practical strategies for handling online services with extremely high request rates—over 200,000 QPS—by avoiding relational databases, employing multi‑level caching, leveraging multithreading, implementing circuit‑breaker and downgrade mechanisms, optimizing I/O, controlling retries, handling edge cases, and logging efficiently.
This extensive tutorial walks through Nginx fundamentals, environment setup, reverse‑proxy load balancing, static‑dynamic separation, resource compression, buffering, proxy caching, IP black‑white listing, anti‑hotlinking, large‑file handling, SSL configuration, high‑availability with Keepalived, and key performance‑tuning techniques for production deployments.
This article examines how moving image pre‑ and post‑processing to GPU with NVIDIA's CV‑CUDA reduces software complexity, alleviates CPU bottlenecks, and delivers up to thirty‑fold throughput gains for computer‑vision workloads across training and inference pipelines.
This article summarizes how Ant Group introduced Alluxio to address storage I/O, capacity, and latency challenges in large‑scale model training, detailing stability improvements through worker‑register follower and master migration, performance gains via follower‑only reads, and horizontal scaling using metadata sharding and multi‑cluster deployment.
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 provides a comprehensive technical overview of Elasticsearch, covering core concepts such as indexes, types, documents, fields, mappings, shards and replicas, the cluster architecture, routing, write and read workflows, segment handling, translog, refresh and flush mechanisms, as well as practical optimization recommendations for hardware, shard strategy, routing, write performance and memory configuration.
This comprehensive guide explains what Elasticsearch does, its underlying Lucene technology, core concepts such as clusters, shards, replicas, mapping, indexing and storage mechanisms, and provides practical performance‑tuning advice for building and operating a robust distributed search engine.
This article explains how moving image preprocessing and post‑processing to GPU with the open‑source CV‑CUDA library dramatically reduces system complexity, eliminates CPU‑GPU bottlenecks, and delivers up to thirty‑fold performance gains for computer‑vision workloads across training and inference stages.
Meituan Visual's engineering team tackled the common low‑GPU‑utilization bottleneck in online inference services by splitting model structures and adopting micro‑service deployment, raising GPU usage from 40% to 100% and more than tripling QPS, and then generalized the approach for other GPU‑based services.
DeWu’s container SRE team combines software‑engineered reliability with routine operations, using defined on‑call roles, SLO/SLA targets, progressive change management, capacity forecasting, four‑metric monitoring, MTTR/MTTF tracking, kernel‑parameter tuning, and namespace‑protected security policies to swiftly resolve incidents such as Redis latency spikes.
Facing massive log volume in JD.com’s shopping cart service, we reduced disk consumption, CPU load, and improved request latency by applying Log4j2 log level filtering, asynchronous logging with AsyncLogger, and TTL‑based thread‑local context propagation, while providing detailed metrics, configuration steps, and best‑practice recommendations.
This article explores practical techniques for optimizing online services handling over 200,000 QPS, covering why relational databases fall short, multi‑level caching, multithreading, circuit breaking, I/O reduction, controlled retries, edge‑case handling, and efficient logging to maintain sub‑300 ms response times.
This article explains common MySQL database performance optimization techniques, underlying storage architecture, index construction and usage guidelines, and presents several real‑world cases from an internet hospital project, helping developers understand optimization strategies, avoid hidden pitfalls, and improve query efficiency.
This article introduces WebAssembly’s efficient, safe, open standards and LLVM‑based compilation pipeline, then guides Rust developers through installing tools, writing and binding simple functions, optimizing performance and bundle size, and explores tooling, runtimes, real‑world use cases, current limitations, and future extensions such as WASI, SIMD, and threads.
This article provides a comprehensive overview of Flutter, covering why to learn it, the architecture layers, widget fundamentals with code examples, Dart language basics, GetX state management, common development pitfalls and solutions, as well as performance optimization techniques for building efficient cross‑platform mobile applications.
This article details Ctrip's experience migrating MySQL workloads to the distributed OceanBase database, covering the design of an assessment tool, a one‑click migration workflow, comprehensive monitoring dashboards, automated fault‑diagnosis pipelines, encountered compatibility issues, and future roadmap for the platform.
The article chronicles Caiyun Weather’s evolution of geographic lookup—from a naïve nearest‑adcode algorithm to a high‑precision H3‑based system—detailing the evaluation of Geohash, S2, and H3, performance benchmarks, memory trade‑offs, and the final R‑Tree‑augmented point‑in‑polygon solution.
This article details Meituan's search‑engine optimization for its food‑delivery platform, describing the performance bottlenecks of Elasticsearch's inverted‑list query and merge phases, the design of a run‑length‑encoding (RLE) index, custom hash‑map term look‑ups, sparse RoaringBitmap structures, integration steps, and the resulting 84% latency reduction.
Facing a four‑hour coupon‑calculation bottleneck at eLong, I explored Storm’s stream‑processing model, identified the data‑extraction stage as the weak link, refactored the hotel‑pull service, leveraged RocketMQ for threading, and ultimately reduced full‑batch processing time to just over an hour.
To support Southeast Asian live-streaming, we designed a custom danmaku system capable of handling up to a million concurrent users per room, tackling bandwidth pressure, weak-network latency, and reliability by employing HTTP compression, response simplification, short-polling, local caching, and lock-free ring buffers.
This extensive tutorial walks through installing Nginx from source, setting up environment, configuring reverse proxy load balancing, static resource handling, compression, buffering, caching, IP black‑white lists, anti‑hotlinking, large file transfer, SSL certificates, high availability with Keepalived, and advanced performance tuning techniques.
Baidu’s Baige AI heterogeneous computing platform delivers an end‑to‑end, low‑cost AI IaaS for autonomous driving, covering data cloud, tiered storage, RapidFS caching, AIAK‑Inference and AIAK‑Training acceleration, GPU container virtualization, and remote GPU pooling, achieving up to 5× faster data access, 391% training speedup, 90% inference latency reduction, and 60% simulation cost cut.
This article curates the ten most-read technical posts from NetEase Yanxuan in 2022, covering DDD adoption, reverse transaction systems, availability metrics, documentation practices, inventory locking, performance optimization, stability governance, H5 containerization, 2B architecture, and the evolution of the Yanxuan app’s engineering architecture.
The article analyzes the growing demands on network I/O, explains Linux’s kernel bottlenecks, and shows how user‑space frameworks like DPDK and UIO can achieve multi‑hundred‑million packets per second through techniques such as huge pages, SIMD, zero‑copy, and careful CPU cache management.
The article walks through a real‑world case of locating the K closest national weather stations to a Beijing site, explains why naive full‑scan using Vincenty's formula is slow, and details a tile‑based indexing solution that cuts query time to 900‑3000 ns and reduces CPU usage by about 10 percent.
Facing memory leaks and high CPU load in a Canvas‑based avatar editor, the team switched to WebGL, leveraging GPU parallelism, off‑screen framebuffers, and blending techniques to dramatically reduce memory consumption and CPU usage while adding features like image layering and color blending.
This article explores the challenges of end‑to‑end tracing on mobile apps, explains why issues are hard to reproduce, and presents a four‑step solution using a unified OpenTelemetry standard, automated data linking, performance optimizations, and machine‑learning‑driven root‑cause analysis.
Ant Group and Hong Kong University of Science and Technology’s award‑winning paper, “Complexity‑guided Container Replacement Synthesis,” presented at OOPSLA 2022, introduces a static‑analysis‑driven method that automatically replaces inefficient Java containers, achieving an average 8.1% runtime improvement across real‑world projects.
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 details Meituan's use of Kafka as a unified data cache and distribution layer, outlines the challenges of massive scale and latency, and presents comprehensive optimizations across application, system, and cluster management layers, including disk balancing, migration acceleration, fetcher isolation, and full‑link monitoring.
stl4go provides a generic Go 1.18 container library that implements an optimized skip‑list‑based ordered map, using adaptive levels, efficient random‑level generation, type‑specific paths, and cache‑friendly node structures to achieve near‑C++ performance, surpassing existing Go generic collections.
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.
This article explains how ICBU measures and improves global front‑end performance using Core Web Vitals and FCP, outlines the rationale behind metric selection, details threshold logic for LCP, CLS and FID, describes multi‑channel data collection and aggregation, and showcases dashboards that help developers identify and resolve performance issues.
This article systematically examines the major performance bottlenecks in AI model training, explains the underlying hardware and software causes, and presents a comprehensive set of acceleration strategies—including data‑loading optimizations, compute‑side enhancements, communication tricks, and the AIAK‑Training suite—backed by real‑world case studies and quantitative results.
This article details the private‑deployment architecture of Volcano Engine A/B Testing (DataTester), outlines three major challenges—version management, performance optimization, and stability—and explains the branch‑logic, release pipeline, model‑optimization, and pre‑aggregation solutions implemented to enable reliable, low‑resource SaaS‑like operation in on‑premise clusters.
This article details Volcano Engine DataTester's private deployment architecture, the version‑management, performance, and stability challenges encountered, and the business‑oriented solutions—including branch strategies, pipeline automation, ClickHouse model optimizations, and multi‑level caching—that enable reliable, efficient A/B testing in on‑premise environments.
The article reflects on developing mobile 3D games, advising developers to choose between lightweight libraries like Three.js and full‑featured engines such as Babylon.js, weigh web, native, or hybrid deployment, adopt appropriate technical trade‑offs, and build a well‑structured 3D team with clear roles to balance performance and visual quality.
This article explains how technical professionals transitioning to management can apply quantitative analysis and systematic measurement to optimize both software system performance—illustrated by an image‑processing pipeline case study—and organizational workflows, highlighting DevOps bottlenecks, resource utilization, and the importance of holistic process improvement.
This article consolidates three performance‑optimization posts, detailing six universal time‑space trade‑off techniques—indexing, caching, compression, prefetching, peak‑valley smoothing, and batch processing—plus four advanced parallelism strategies, illustrated with real‑world analogies and practical guidelines for developers.
This article examines the evolving demands of network I/O, analyzes Linux and x86 bottlenecks, explains DPDK's user‑space bypass architecture, and presents practical optimization techniques—including huge pages, SIMD, cache prefetching, and compile‑time tricks—to achieve multi‑gigabit packet processing rates.
By replacing the default JavaScriptCore engine with Hermes for Hippy’s UI rendering in QQ Browser, the team cut module load time by 70‑80 %, lowered first‑frame latency and crash rate, kept memory usage comparable, and leveraged Hermes’s bytecode support and React‑Native ecosystem to dramatically improve startup performance.
This article explains how to batch multiple cluster endpoint requests in Go by using a token‑based queue and concurrent workers, dramatically reducing load on the registration center while requiring minimal changes to existing code.
This article systematically analyzes the main performance bottlenecks in AI model training, explains why acceleration is essential, and presents current hardware‑ and software‑based solutions—including data‑loading optimizations, operator fusion, mixed‑precision and Tensor Core usage, as well as distributed communication strategies—followed by real‑world case studies of Baidu's AIAK‑Training suite that demonstrate significant speed‑ups.
This article explains the architecture of Weibo's popular recommendation system, the role of the weidl online learning platform, and how DeepRec’s performance optimizations—such as oneDNN operator acceleration, cost‑aware scheduling, and adaptive memory allocation—significantly improve training speed, inference latency, and overall service throughput.
This article recounts a year‑long, hands‑on experience of deploying and continuously optimizing Apache SkyWalking for full‑link monitoring in a large micro‑service environment, covering the motivations, architecture choices, pre‑research, POC integration, and a series of performance‑tuning steps that reduced segment storage from billions to millisecond‑level query latency.
This article describes the background, technical selection, architecture, deployment, performance tuning, and operational practices of a Milvus‑driven vector retrieval platform, including cloud‑native deployment, index choices, capacity planning, and real‑world application cases that improve recall latency and resource efficiency.
This article introduces a two-level cache framework designed to address issues with Spring-Cache, providing enhanced caching strategies and improved performance in Java-based backend systems.
This article shares Ctrip's practical experience in microservice governance, detailing the background, common pitfalls such as excessive service granularity and cyclic dependencies, and presenting concrete goals, principles, and strategies that led to significant improvements in stability, performance, and development efficiency.
By eliminating unnecessary reverse‑geocode calls, aggregating nearby coordinates with GeoHash, and employing a multi‑layer LRU‑K cache with time‑ and access‑count eviction, the Hellobike map team cut daily requests from 200‑300 million to 20‑30 million while adding fallback and monitoring mechanisms.
This article explains React's internal update loop, the render and commit phases, default bailout optimizations, and how developers can accelerate rendering using techniques such as state lifting, content lifting, React.memo, useMemo, and useCallback while avoiding over‑optimization.
Bilibili's cloud‑based Android build system leverages high‑performance remote servers, Dockerized build images, Git diff synchronization, and distributed compilation of heavy tasks like DexBuild to dramatically accelerate builds for its massive monorepo, providing parallelism and VIP priority while incurring a learning curve and added maintenance costs.
Tencent Cloud’s OCR team cut average response time from 1.8 seconds to under one second and boosted throughput by over 50 % by redesigning the model with self‑attention, accelerating inference with a Tensor‑Network accelerator, shrinking RPC payloads, enabling asynchronous logging, and optimizing multi‑region GPU memory utilization.
The article details vivo's development of a high‑accuracy, high‑performance streaming speech‑recognition inference engine built on the wenet framework, its optimization techniques such as dynamic batching and memory pooling, collaborative acceleration with KunlunChip's high‑performance inference library, and extensive performance benchmarks demonstrating multi‑batch GPU and XPU gains.
The team created Layout2Code, an APT‑based generator that translates XML layouts into Kotlin code at build time—handling merge and include tags and supporting custom Views—cutting homepage layout inflation by over 200 ms, boosting performance metrics while requiring far less maintenance than traditional X2C solutions.
This article describes how Qunar’s travel app team built a unified performance measurement platform and applied a series of frontend‑focused optimizations—including JSBundle pre‑loading, slimming, splitting, rendering‑stage improvements, and service‑request prefetching—to consistently reduce the Time‑to‑Interactive metric by double‑digit percentages across core booking pages.
This article examines the selection of ClickHouse as the OLAP engine for a self‑service analytics platform, describes the platform’s architecture, details memory and performance tuning techniques, discusses large‑scale join handling, and outlines current challenges and future development directions for ClickHouse.
In early 2021 senior engineer Liu Lei led a complete redesign of Bilibili’s PC homepage, adopting Vue 3 with a custom SSR/CSR hybrid, responsive grid layouts via CSS grid and flex, theme variables, monorepo tooling, and extensive performance tweaks—boosting the page’s score dramatically while preserving hackability and supporting both narrow and ultra‑wide screens.
The article proposes a full‑stack H5 containerization framework for mobile apps that boosts user experience and development efficiency by pre‑heating WebViews, using offline packages and parallel data loading to cut load times, adding comprehensive performance monitoring, a modular native‑plugin system, and a configuration‑driven shell‑container factory that frees frontend teams from mobile‑team dependence.
This article explains how Alibaba Cloud EMR StarRocks extends data lake analytics to support Hive, Iceberg, and Hudi, detailing its architecture, Iceberg integration, performance gains over Trino, IO merging, lazy materialization, intelligent caching, and elastic compute capabilities for faster, unified, and cost‑effective queries.
This article explains the concept of Redis data skew, its causes and impacts, explores data volume and access skew classifications, presents mitigation strategies, and provides a comprehensive source‑code walkthrough of JD's open‑source hotkey framework—including client, worker, and dashboard components—for detecting and handling hot keys in distributed cache clusters.
This article details Meituan's large‑scale Kafka deployment—over 15,000 machines and petabyte‑level daily traffic—its operational challenges such as slow nodes, load imbalance, and resource contention, and the comprehensive read/write latency, system‑level, and cluster‑management optimizations implemented to improve performance and reliability.
The article surveys GPU rendering pipelines and hardware architectures for desktop and mobile, explains classic stages, compares Immediate Mode, Tile‑Based and Tile‑Based Deferred rendering, details PowerVR, Mali and Adreno components, and offers optimization advice on draw calls, depth pre‑passes, shader efficiency, and render ordering.
This curated guide explores key front-end topics—including unified browser storage APIs, comprehensive engineering configuration practices, deep dives into exception handling and performance optimization, essential packet-capture techniques, impactful micro-copy design, and a personal four-year front-end development journey—providing practical insights for modern web developers.
This article describes the architecture, key features, dynamic deployment, performance optimizations, and real‑world results of a high‑throughput online inference platform that serves deep‑learning models for JD.com’s risk‑control decision engine, achieving near‑hundred‑fold latency improvements.
An accidental discovery by a GitHub engineer revealed that moving a single '&' operator in a Go program eliminated unnecessary copies, reduced heap allocations via escape analysis, and improved execution speed by 42%, highlighting common pitfalls in Go’s regex engine, malloc, and garbage collection.
This article introduces front‑end lazy‑loading strategies—including route lazy loading, three image lazy‑loading methods (native, scroll‑event, and IntersectionObserver), and module lazy loading—explaining their principles, showing complete React/Vue code examples, and demonstrating how they reduce initial load time and improve user experience.
UCloud’s US3 object storage metadata service, originally built on a chained MongoDB architecture, faced scalability, performance, and cost challenges, prompting a redesign that introduces a high‑compatibility DB‑Gateway, a distributed KV store (UKV) with custom RocksDB, delivering faster reads, zero list‑service latency, 80% cost reduction, and simpler operations.
This article explains how to build a reliable e‑commerce inventory system by handling concurrent stock deductions, designing inventory and flow tables, using optimistic locking and transactions, and scaling the solution with read‑write splitting and Redis caching to meet high‑QPS demands.
Meituan's food‑delivery search team identified heavy CPU and latency hotspots in Elasticsearch's posting‑list query and merge phases, then redesigned the inverted index using Run‑Length Encoding, hash‑based term lookup, index sorting and a custom SparseRoaringDocIdSet, ultimately reducing TP99 search latency by 84% and cutting CPU usage dramatically.
This article reviews a decade of Google TPU development, highlighting ten technical and architectural lessons, the hardware's impact on the AI industry, performance and energy‑efficiency improvements, and strategies for reducing machine‑learning carbon footprints.
Uber processes over 500,000 daily Presto queries across 20 clusters handling more than 50 PB of data, and by deploying Alluxio Local Cache on NVMe disks they raised cache‑hit rates from roughly 65% to over 90% while addressing real‑time partition updates, node churn, and cache‑size constraints.
This article explains the value, underlying HTTP redirect mechanism, and detailed design approaches—including hash‑based and distributed‑ID generation, base‑62 encoding, database storage, caching, indexing, sharding, and rate‑limiting—to build a scalable short‑link service.
This guide details practical performance‑optimization techniques for Cocos‑based WeChat mini‑games, covering scene and prefab optimizations, asset compression, code size reduction, garbage‑collection handling, mask usage, plugin assistance, code‑level improvements, and a suite of testing tools to measure and verify the gains.
