Bilibili's new AI Infinite Arena evaluates large models through real-world tasks by content creators, with GPT-6 Astra leading the first round over GLM-5.3 and Chinese models taking three of top five spots, though rankings reflect specific scenarios not universal superiority.
The article examines Soulter’s early‑2022 launch of AstrBot, its growth into a multi‑platform agent orchestration framework, compares it with the more consumer‑friendly OpenClaw, and analyzes how Bilibili’s unique feedback loop amplifies AI projects and shapes the Chinese AI creator ecosystem.
This article analyzes Bilibili's private messaging architecture, identifies performance bottlenecks such as MySQL slow queries, storage limits, and tightly coupled services, and presents a multi‑layer redesign with caching, sharding, BFF separation, and consistency mechanisms to sustain tenfold traffic growth.
This article explains how to replicate Bilibili's subtitle masking effect by using a single mask image and the CSS mask‑image property, providing a complete HTML/CSS demo, visual results, and cautions about its experimental status.
This article explains Bilibili's large‑scale data platform architecture, the common offline‑task failures and slowdowns users encounter, and how a large language model‑driven intelligent assistant is being built to automatically diagnose and resolve these engineering problems.
This article explains Bilibili's large‑language‑model‑based intelligent assistant, describing the platform's five‑layer architecture, massive daily task load, common failure and slowdown causes, and the need for AI‑driven troubleshooting to improve reliability and performance.
Discover how to transform Claude AI into a powerful research assistant that automatically fetches Bilibili video subtitles, converts them into clean markdown, analyzes the content, and produces structured product selection reports, all without manually watching videos, using Python scripts, cookies, and Claude Skills.
This article explains how a single image and the CSS -webkit-mask-image property can be used to prevent subtitles from covering on‑screen characters, replicating Bilibili's non‑obstructive bullet subtitles with a lightweight, experimental technique demonstrated through a complete HTML demo.
This article explains Bilibili's massive data platform architecture, the common offline‑task failures and slowdowns users encounter, and how the company applies a large‑language‑model‑driven intelligent assistant to diagnose and resolve these engineering problems efficiently.
Exploring Bilibili’s large‑scale data platform, this article details its five‑layer, storage‑compute separated architecture, the massive daily workload of offline and real‑time tasks, common failure and slowdown causes, and how an LLM‑powered intelligent assistant is being developed to help engineers troubleshoot efficiently.
This article explains Bilibili's massive big‑data platform architecture, the common offline‑task failures and slowdowns users encounter, and how a large‑language‑model‑driven intelligent assistant is being built to diagnose and resolve these issues automatically.
This article explains how Bilibili's massive video platform leverages a large‑language‑model‑driven assistant to diagnose and resolve offline task failures and slowdowns within its five‑layer, storage‑compute separated big data architecture, improving operational efficiency for thousands of daily queries.
This article explains Bilibili's massive data platform architecture, the common reasons offline tasks fail or slow down, and how the company is exploring large‑language‑model‑driven assistants to automatically diagnose and resolve these engineering issues.
This article explains how a simple CSS mask‑image technique can make on‑screen subtitles avoid covering characters, reproducing Bilibili's non‑obstructive caption effect with just an image and a single CSS property, complete with a working demo and usage cautions.
This article details Bilibili's implementation of a large‑language‑model‑driven intelligent assistant that helps engineers diagnose and resolve massive offline and real‑time data‑processing failures, describing the platform’s five‑layer architecture, common failure and slowdown causes, and the need for AI‑powered troubleshooting support.
This article explains Bilibili's massive video platform data architecture, the huge daily workload of offline and real‑time tasks, common user problems like task failures and slowdowns, their root causes, and how a large language model assistant is being used to automate troubleshooting.
This article explains Bilibili's large‑scale big‑data platform architecture, the huge volume of offline and real‑time tasks it handles, common failure and slowdown causes, and why the company is exploring AI‑driven assistants to help engineers troubleshoot these issues efficiently.
Bilibili’s massive video platform relies on a five‑layer, storage‑compute separated big‑data architecture, handling hundreds of thousands of daily tasks, and now leverages large‑language‑model assistants to automatically diagnose and resolve frequent task failures and performance slowdowns.
This article explains Bilibili's implementation of a large‑language‑model based intelligent assistant, detailing the platform's five‑layer architecture, the huge volume of offline and real‑time jobs, common user issues like task failures and slowdowns, and how AI can help automate troubleshooting.
This article explains how Bilibili leverages a large‑language‑model‑based intelligent agent to diagnose and resolve failures and slowdowns in its massive big‑data platform, detailing the platform architecture, workload scale, common user issues, and the need for automated assistance.
This article explains how Bilibili leverages a large‑language‑model‑driven assistant to diagnose and resolve failures and slowdowns in its massive big‑data platform, detailing the platform’s five‑layer architecture, common task issues, and the need for intelligent troubleshooting tools.
This guide explains how to set up the PHP‑Bilibili‑Danmu framework built on Webman, configure Protocol Buffers, run database migrations, connect to Bilibili live rooms, handle danmu and interaction events, and verify the installation using a management dashboard.
This article details Bilibili's third‑generation streaming transcoding system, explaining its architecture, core modules, testing strategies, quality‑assurance plans, and code examples that together dramatically reduce processing time and support millions of daily video uploads.
This tutorial walks you through building a Python project that downloads a Bilibili video, extracts its frames, applies Baidu AI human segmentation, scrapes danmu comments, generates a stylized word‑cloud animation, and finally composes a video with background music, showcasing video processing, AI, and data visualization techniques.
Bilibili performed client-side DCDN disaster-recovery drills using a self-built HTTPDNS to simulate DNS, CDN, and SSL faults; automated scripts across Android, iOS, and Web injected errors, measured rendering latency, validated immediate downgrade to commercial services, refined fallback strategies, and demonstrated near-zero user impact during a real network incident.
The paper presents Bilibili’s multi‑layer CDN edge‑node scheduling system, which groups heterogeneous nodes by quality and price, uses cost‑aware and resource‑aware heuristics—including maximum‑flow regional borrowing and contextual‑bandit utilization prediction—to allocate bandwidth per business, achieving a 43 % bandwidth reuse increase, 33 % coverage boost, and markedly lower stall rates.
A former Bilibili front‑end engineer inserted malicious code that displays a fake ban screen, deletes users' uploaded videos at high speed, and reveals how insider threats can turn a popular video platform into a security nightmare.
A former Bilibili engineer inserted malicious JavaScript into the platform's web frontend, causing some users to see a fake account‑ban notice, which exposed internal code‑review weaknesses and sparked widespread discussion about insider threats and web security.
We built an LLM‑powered system for Bilibili that automatically creates ad titles from user keywords, employing fluency, style, and quality classifiers, mixed domain data cleaning, and alignment methods such as SFT, DPO and KTO, resulting in a product that now generates about ten percent of daily titles and drives significant ad spend.
Bilibili’s Activity Platform low‑code system was completely redesigned over two years, introducing a modular, decoupled architecture with an iframe‑based canvas, unified renderer, and Eva‑CLI tools that cut activity build time from three days to 0.48 days, boosted coverage to 77% of activities, and reduced first‑contentful‑paint by 35%.
This tutorial explains how to capture Bilibili video streams with Fiddler, extract the necessary request headers and URLs, and use a concise Python script based on the requests library to download videos, handling header nuances and batch‑fetching video lists from an uploader’s page.
Bilibili’s large‑scale recall system separates online processing into a two‑tier merge service and an index service, supports multi‑channel text, item‑to‑item and vector indexes with real‑time updates, uses horizontal sharding, robust CI/CD, monitoring and degradation mechanisms, and is being extended toward model‑based recall and greater automation.
Bilibili replaced a restrictive third‑party video editor with a self‑developed engine, redesigning architecture for extensibility, manageability and controllable rollout, refactoring hundreds of API calls, enabling draft migration, adding observability, and achieving lower crash rates, faster timelines and stable conversion gains while continuing AI‑assisted feature expansion.
This article details Bilibili's end‑to‑end front‑end error monitoring solution, covering the custom SDK, error capture and classification, unique ID generation, filtering, white‑screen detection, data pipelines, APM visualisation, lifecycle plugins, one‑click alerts, and future roadmap, all backed by real‑world metrics and code examples.
This article demonstrates how to recreate Bilibili's subtitle effect that avoids covering on‑screen characters by using a simple HTML structure combined with the CSS mask‑image property, providing a complete demo, visual results, and practical considerations for frontend developers.
Bilibili transformed its rapidly growing charging‑exclusive video and short‑drama services into a modular, platform‑based architecture that separates manuscript, video, charging and drama domains, adds fine‑grained data operations, real‑time observability and robust stability mechanisms, thereby reducing technical debt and enabling scalable, customizable creator‑fan interactions.
Bilibili tackled race‑condition vulnerabilities in its massive video content pipeline by replacing simple timestamp checks with optimistic locking (CAS) and rate‑limiting locks, adding version verification and observation tools that now eliminate missed reviews and improve security, scalability, and real‑time editing reliability.
Bilibili’s game live‑stream platform implements interactive features ‘Play Together’ and ‘Help Me Play’ by using Redis ZSET queues, MySQL persistence, real‑time streamer recommendation, ticket‑based purchase flows, state‑machine order handling, and comprehensive monitoring to ensure reliable, scalable viewer‑streamer gameplay collaboration.
The paper presents a scalable, message‑queue‑driven reward system for Bilibili Live that unifies diverse reward types and distribution scenarios through standardized APIs, layered fast/slow queues, idempotent processing, multi‑stage retries, and comprehensive monitoring to ensure low latency, no over‑issuance, and reliable delivery.
This article details Bilibili's transformation of its search offline indexing pipeline, moving from manual MySQL‑based processes to a high‑capacity, distributed KV store and Spark‑driven builds, addressing performance, maintenance, and scalability challenges while improving resource efficiency and iteration speed.
Bilibili’s Change Prevention Platform consolidates data from over 60 systems to proactively detect and block more than 100 risky changes daily, reducing change‑related incidents by applying a four‑pillar framework of technical support, landing, cross‑domain enablement, and cultural safeguards, while evolving toward AI‑driven, end‑to‑end change defense.
This article analyzes Bilibili's creator service ecosystem, detailing the lifecycle‑based support model, industrial‑scale creation tools, copyright protection mechanisms, growth incentives, and data‑driven feedback systems that together aim to boost creator efficiency, safety, and long‑term success on the platform.
At WAIC 2024, Bilibili unveiled a suite of AI tools—including a 1:1 digital‑avatar generator, dynamic comic technology, a customized voice library for virtual singer Luo Tianyi, the BiliStudio video‑audio model, and the Index‑1.9B large‑language models—empowering creators, cutting production costs, and serving its 80 million‑plus monthly users with advanced content‑creation and commercial‑marketing solutions.
Bilibili’s video submission platform uses a layered micro‑service architecture with a DAG‑based scheduler, extensive observability, and HA tactics such as sharding, 64‑bit ID migration, full‑link stress tests, chaos engineering, and multi‑active data‑center deployment, while tooling like trace correlation and automated alerts ensures stability and guides future hybrid‑cloud migration.
This article outlines Bilibili's data‑driven approach, describing the five‑year development of its big‑data development governance platform, its user segmentation, product positioning, data‑map and governance product designs, operational methods, value evaluation, and future roadmap, highlighting significant efficiency gains and user impact.
At the 2024 GOPS Global Operations Conference, Bilibili senior R&D engineer Gu Lintao will present BCM—Bilibili’s Chaos Engineering Platform—showcasing how its design and capabilities let developers, testers, and SREs safely inject faults, uncover hidden architectural risks, and improve service stability through real‑world drills and systematic reliability engineering.
Bilibili has finished beta development of its HarmonyOS native video app—the first bullet‑comment‑enabled video client on the platform—offering medium‑long video playback, personalized recommendations, search, hot lists and interactive features, while cutting build times from over twenty to under five minutes and targeting a full commercial launch with expanded services across smartphones, tablets, displays and vehicles in Q4 2024.
Bilibili’s self‑supervised video super‑resolution system upgrades low‑resolution streams to 4K by using three parallel degradation‑branch networks—texture‑enhancing, line‑recovering, and noise‑removing—tailored to anime, game, and real‑world content, delivering sharper edges, finer textures, and measurable quality gains across its online playback pipeline.
Unmanned live streaming—pre‑recorded video played continuously—can be achieved by capturing an MP4 file and using FFmpeg to push it to an RTMP endpoint, with step‑by‑step instructions for Bilibili and WeChat Video channels, including RTMP URL acquisition, command syntax, and verification.
Bilibili combats personal‑attack and trolling comments by combining sector‑specific keyword databases, user‑group analysis, advanced word‑matching (including pinyin and homophone detection) and multiple NLP/graph models, which has cut personal‑attack reports in entertainment, film and gaming by about 32 % and trolling reports by roughly 25 % between June and December 2023.
BilibiliContent moderationabusive language detection
This article details Bilibili's five‑year development of a comprehensive big‑data governance platform, covering its usage scenarios, product positioning, data map and governance solutions, abstract configuration approach, operational mechanisms, and future plans, highlighting significant improvements in data efficiency and value assessment.
Bilibili’s VOD player cuts bandwidth waste and costs by up to 15% through an intelligent caching and pre‑loading system that evaluates network risk and user habits to size cache dynamically, reduces stalls by 20% in vertical mode, and is continuously refined via a multi‑layer monitoring framework.
Bilibili’s hybrid‑cloud architecture combines multiple public clouds and its own data centers using a star‑centered topology, employing VPCs, load balancers, NAT gateways, and dedicated lines managed via a multi‑cloud platform, while outlining project‑network setup, security rules, and proxy‑forwarding solutions for reliable, cost‑effective cloud networking.
The article explains how a simple CSS mask‑image technique can replicate Bilibili's subtitle behavior that avoids covering on‑screen characters, providing a minimal HTML/CSS demo, visual results, and a caution about the experimental nature of the property.
The article details Bilibili's comprehensive anti‑crawling solution, covering the problem background, a two‑layer detection framework integrated with APIGW and GAIA, risk perception, strategy iteration, verification mechanisms, quantitative results, and future improvement directions, all illustrated with concrete examples and performance numbers.
This article details Bilibili's comprehensive anti‑crawling system, covering the background of API abuse, the data‑flow framework, risk perception, strategy iteration, verification mechanisms, gateway signing design, and the measurable impact on normal and special‑case interfaces.
Bilibili combats API abuse by deploying a two‑layer anti‑crawling system—gateway‑side signature verification and a GAIA risk‑control engine integrated into APIGW—that unifies device data, applies flexible rule packages, triggers diverse human challenges, and has already blocked billions of malicious requests with over 85% recall while preventing service outages.
This article details Bilibili's step‑by‑step transformation of a fragmented, high‑coupling live‑commerce system into a modular, platform‑centric architecture, covering product middle‑platform construction, unified standards, storage migration, monitoring with Prometheus/Grafana, and performance gains such as a three‑fold query speedup and a reduction of development cycles from 46 to 5 person‑days.
Bilibili revamped its alert monitoring platform to meet rapid growth, focusing on effectiveness, timeliness, and coverage; it introduced a closed‑loop design and governance that cut weekly alerts by 90%, built a knowledge‑graph root‑cause system achieving 87.9% accuracy with sub‑minute latency, and integrated AIOps for ongoing refinement.
Bilibili transformed its fast‑growing live‑streaming e‑commerce operation by constructing a modular platform that separates product, user, and application layers, introduces a unified product middle‑platform, standardized capabilities, real‑time attribute handling, and robust monitoring and governance, thereby reducing technical debt, improving stability, and preparing for hundred‑billion‑level GMV scaling.
This article details Bilibili's comprehensive approach to standardizing event tracking (埋点), covering its definition, data pipeline, common business issues, metadata‑driven management strategies, efficiency gains, and future prospects for unified real‑time and batch processing.
This article details Bilibili's data quality assurance framework, covering its evolution across four data platform stages, the architecture of its quality data warehouse, core capabilities such as a complete assurance system, digital‑driven continuous optimization, and efficient incident handling, plus case studies, future plans, and a Q&A session.
Bilibili's SQLScan is a static‑code analysis tool that parses Hive, Spark, Presto and Flink SQL via Antlr4, builds a unified AST, applies engine‑specific metadata plugins for rule enforcement, provides field‑lineage and cost‑analysis services, and has processed hundreds of thousands of daily queries, intercepting thousands of problematic statements to improve data quality and operational efficiency.
This article details Bilibili's offline big‑data cluster challenges, the mixed‑deployment architecture that combines offline and online resources, the Amiya service's over‑commit and eviction mechanisms, performance optimizations, monitoring strategies, and future plans to further improve resource utilization and scheduling.
This article details the evolution of Bilibili's dynamic feed system, outlining its business characteristics, multi-layer microservice architecture, caching strategies, and the transition from incremental pull to a hybrid push‑pull model to address scalability, latency, and data consistency challenges.
This article walks through discovering Bilibili's comment API, programmatically fetching paginated JSON data, converting it into Java POJOs, storing and sorting the comments, and finally feeding the top entries to ChatGPT for automated sentiment and content analysis.
The article traces Bilibili’s Dynamic feed evolution from a legacy client‑server model to a four‑layer micro‑service architecture, detailing cache‑list storage, sharded relational and KV databases, and a push‑pull hybrid feed pipeline that balances massive write traffic, read efficiency, and future interest‑based recommendations.
Bilibili’s FinOps initiative transformed its hybrid‑cloud environment by establishing cross‑department collaboration, deploying a multi‑cloud management platform, and leveraging data‑driven budgeting and vendor selection, enabling systematic detection of waste, precise resource attribution, and continuous cost‑reduction while supporting sustainable, low‑carbon operations.
By completely rebuilding its topic system in 2021, Bilibili introduced an AI‑driven pipeline that automatically discovers, creates, ranks, and populates hot and cold‑start topics using real‑time metrics, rule‑based and vector‑based recall, dramatically boosting content relevance, user interaction, and operational efficiency across the platform.
This article explains Bilibili's comprehensive approach to event‑tracking (埋点) standardization, covering the definition, data pipeline, common business issues, metadata‑driven design strategies, efficiency gains in accuracy, storage and querying, and future directions for automated data flow.
Bilibili’s Q4 2022 initiative overhauled the overseas video upload flow by defining new metrics, adding progress indicators, shrinking chunk size, enabling multi‑threaded and resumable uploads with background support, which boosted submission conversion by up to 10 % and cut failure rates from 4 % to 1 %.
Bilibili’s Cloud Multimedia team uses AI‑driven pipelines to cut bandwidth costs while delivering low‑bitrate, high‑quality video, employing a QoE‑based decision engine, real‑time 4K super‑resolution for game streams, low‑rank reconstruction for narrow‑band HD, data‑driven HDR LUTs, and explores diffusion‑based restoration for legacy content.
This article details Bilibili's SRE team's design and implementation of a high‑availability multi‑active architecture, covering zone types, same‑city and cross‑region deployments, traffic routing, cache consistency, message handling, governance, and practical lessons learned from real‑world incidents.
Amiya, a self‑developed dynamic over‑commit component for Bilibili’s offline big‑data cluster, inflates reported resources on under‑utilized nodes and adjusts them when load rises, adding roughly 683 TB of memory and 137 k vCores, boosting per‑node memory by 15 % and CPU usage by over 20 % while keeping eviction rates below 3 %.
A recent Bilibili patent outlines a method for converting spoken commands into bullet‑screen comments within car terminals, aiming to enable hands‑free video interaction while sparking debate over its practicality and future adoption.
The article explains Domain‑Driven Design as a method for managing software complexity, outlines its meta‑models and limitations, and demonstrates a full problem‑domain analysis for Bilibili’s OGV business—including stakeholder value‑demand analysis, process and use‑case modeling, scenario definition, ubiquitous language creation, and sub‑domain classification—while noting that implementation details will be covered later.
Bilibili’s data platform upgraded its dependency model by shifting from project‑level to task‑level dependencies, adding root and end nodes, using virtual tasks for external data, introducing offset handling, implementing an abstract DependencySubject and asynchronous callbacks, achieving sub‑second latency for tens of thousands of daily tasks while planning automated lineage and richer rule support.
This article explains how Bilibili’s KV storage system evolved from early Redis/Memcache solutions to a custom distributed KV architecture, detailing design goals, architecture components, shard management, Raft replication, multi‑active disaster recovery, and the operational challenges solved to support massive traffic growth.
This article describes the design, algorithmic improvements, and engineering performance optimizations of Bilibili's massive video duplicate detection (collision) system, covering challenges of low‑edit‑degree reposts, two‑stage retrieval, self‑supervised feature extraction, GPU‑accelerated preprocessing, and the resulting gains in accuracy and throughput.
On World Intellectual Property Day 2023, Bilibili highlighted ten award‑winning patents—including on‑screen creator notes, personalized danmaku recommendation, video hotness forecasting, live‑stream highlight detection, mobile runtime observability, audio‑video sync measurement, automated script generation, occlusion‑aware comments, cloud‑gaming force feedback, and synchronized playback timestamps—showcasing its commitment to technical innovation.
This article details Bilibili's evolution of big data governance, describing the early data growth challenges, the launch of the "Wanglou" project, the development of asset metadata and governance indicator frameworks, storage cost reduction strategies, scoring models, and the shift from passive, single‑point fixes to proactive, multi‑dimensional governance across the organization.
Bilibili’s exabyte‑scale big‑data platform, after rapid growth created fragmented ownership and costly storage, launched the Wanglou project to build a metadata‑driven, indicator‑based governance framework that cut storage use by half, introduced compliance scoring and automation, and now plans to extend proactive, multi‑dimensional governance to compute, traffic and lake‑house resources.
This article presents an in‑depth overview of Bilibili’s big data development governance platform, detailing its architecture, the pain points of platform construction, data‑driven methodology, product strategy, and practical solutions for data integration, quality, cost, and security governance across large‑scale data operations.
Bilibili's capacity management platform integrates foundational data, VPA/HPA scaling, quota control, and visual dashboards to streamline resource usage, cut costs, and boost stability, delivering event‑specific support such as for S12 that slashes release issues by 80% and online failures by 90%, while planning predictive scaling and risk control.
At Bilibili, extensive research showed AVIF can cut thumbnail size by about 35% versus WebP, prompting a layered implementation that uses libheif/libaom encoding with fallback WebP, client‑side libavif decoding, comprehensive monitoring, and solutions for encoding latency, while planning hardware acceleration and continued standard‑setting contributions.
Bilibili’s high‑quality ASR system combines large‑scale filtered business data, semi‑supervised Noisy‑Student training, an end‑to‑end CTC model with lattice‑free MMI decoding, and FP16‑optimized FasterTransformer inference on Triton, delivering top‑ranked accuracy, low latency, and scalable deployment for diverse Chinese‑English video content.
The article details Bilibili’s thumb‑up service architecture, covering required business and platform capabilities, handling high read/write traffic and hot‑spot pressures, a three‑tier storage design using TiDB, Redis cache and local heap cache, disaster‑recovery mechanisms, asynchronous jobs, and future modularization plans.
This article presents a comprehensive overview of Bilibili's comment system, detailing its background, core functional modules, layered architecture, storage and caching strategies, availability considerations, and hot‑comment design, illustrating how the platform achieves scalability, performance, and reliability for massive user interactions.
This article presents a detailed study on applying recommendation‑system techniques to test case prioritization for Bilibili's mobile apps, describing the problem definition, evaluation metrics, data processing, FM model selection, experimental results, practical deployment, and future research directions.
BilibiliTest Case Prioritizationfactorization machine
This article explains how to replicate Bilibili’s non‑overlapping subtitle effect using a simple HTML demo and the experimental CSS mask‑image property, detailing the implementation steps, code example, and visual results while noting its experimental status.
To support Bilibili’s 330 million MAU growth, the network team replaced the static, single‑core 1.0 public‑network with a regionalized, SLA‑driven 2.0 architecture featuring centralized POPs, dual‑arm LB/NAT, clustered services, security gateways, and a multi‑level CDN backbone that reduces latency, improves reliability, and cuts bandwidth costs.
This article explains how a single mask image and the CSS mask-image property can be used to prevent subtitles from overlapping on‑screen characters in Bilibili videos, providing a lightweight demo with HTML/CSS code, visual results, and practical cautions about browser support.
The article details how Bilibili partnered with Huawei Cloud to stream the 2022 League of Legends World Championship, using real‑time super‑resolution, microservice architecture, edge computing and interactive features like heat effects and shopping carts, delivering a seamless, high‑quality, immersive esports experience for hundreds of millions of viewers.
To meet Bilibili’s demanding live and on‑demand video transcoding needs, the BANG engine combines a multi‑stage pipeline architecture, frame‑block and multi‑frame parallelism, SIMD‑based CPU acceleration, and TensorRT/TensorFlow GPU inference, offering configurable string‑based pipelines that dramatically increase throughput while simplifying integration.
Bilibili built a browser‑native live P2P streaming system that uses WebRTC data channels to exchange 60 KB‑sized HLS segment blocks via a WebSocket tracker, employs MessagePack for efficient binary messaging, and adopts a decentralized free‑market peer‑role allocation to limit uploads, dramatically cutting bandwidth while supporting massive concurrent viewers.
This article chronicles Bilibili Live's eight‑year journey from a simple LAMP monolith to a sophisticated microservice architecture, detailing the migration to Swoole‑based services, Docker containerization, Golang adoption, hot‑key handling, request amplification mitigation, and the creation of a high‑performance Envoy‑based gateway.
This article explains how a single image and the CSS mask‑image property can prevent subtitles from covering on‑screen characters, provides a complete demo with code, shows the visual result, and discusses the experimental nature and limitations of the technique.
The article details Bilibili’s capacity‑management approach, showcasing system architecture, monitoring metrics, scaling tactics, and performance results through diagrams and screenshots, and explains how its operational processes and tools maintain reliable service delivery during high‑traffic periods, offering practical capacity‑planning insights.