2025-12-03 Tech Briefing | Uber uses MGR to enhance MySQL cluster availability.

Wednesday

Welcome to this edition, featuring 24 articles.

📚 Contents/📝 TL;DR

📚 Contents

📝 TL;DR

Enhancing MySQL Cluster Availability: Designing Advanced Detection, Mitigation, and Consensus Mechanisms with Group Replication

Uber uses MGR to enhance MySQL cluster availability.

Enhancing MySQL Cluster Availability: Designing Advanced Detection, Mitigation, and Consensus Mechanisms with Group Replication

Uber uses MGR to enhance MySQL cluster availability.

Masonry Layout: The Future Where You No Longer Need Libraries

Web platform native features replace JS libraries, enhancing frontend performance.

Masonry Layout: The Future Where You No Longer Need Libraries

Web platform native features replace JS libraries, enhancing frontend performance.

Colleague: The architecture is too complicated, I spent hours looking for the source code file. Me: Have you tried source code tracing?

Vite plugin combined with Greasemonkey script to achieve fast source code location in frontend.

Colleague: The architecture is too complicated, I spent hours looking for the source code file. Me: Have you tried source code tracing?

Vite plugin combined with Greasemonkey script to achieve fast source code location in frontend.

Should Tailwind CSS Bear the Blame for Writing a Messy Mountain of Code?

A senior frontend engineer shares their experience and practices of transitioning from resistance to appreciation of Tailwind CSS.

Should Tailwind CSS Bear the Blame for Writing a Messy Mountain of Code?

A senior frontend engineer shares their experience and practices of transitioning from resistance to appreciation of Tailwind CSS.

Choosing Between Flutter and Electron for Desktop Application Development

In-depth comparison of Flutter and Electron for desktop application development.

Choosing Between Flutter and Electron for Desktop Application Development

In-depth comparison of Flutter and Electron for desktop application development.

Masonry Layout: When You No Longer Need Libraries

Native Web Platform Features Replace JS Libraries, Enhancing Frontend Performance

Masonry Layout: When You No Longer Need Libraries

Native Web Platform Features Replace JS Libraries, Enhancing Frontend Performance

Modi Factory 022 Week AI Application Ranking: PowerMem Leads with Long-Term Memory AI Infrastructure, This Week AI Applications Embrace the Trend of 'Long-Termism'

PowerMem leads the AI application rankings, focusing on long-term memory and infrastructure.

Modi Factory 022 Week AI Application Ranking: PowerMem Leads with Long-Term Memory AI Infrastructure, This Week AI Applications Embrace the Trend of 'Long-Termism'

PowerMem leads the AI application rankings, focusing on long-term memory and infrastructure.

Five Challenges of Enterprise AI Adoption, Lenovo Lingtuo Upgrades Storage Product Matrix

Lenovo Lingtuo releases AI storage solutions to address enterprise AI adoption challenges.

Five Challenges of Enterprise AI Adoption, Lenovo Lingtuo Upgrades Storage Product Matrix

Lenovo Lingtuo releases AI storage solutions to address enterprise AI adoption challenges.

Azure Virtual Desktop Achieves Full Hybrid Deployment via Arc-Enabled Servers

AVD enables on-premises hybrid deployment via Arc

Azure Virtual Desktop Achieves Full Hybrid Deployment via Arc-Enabled Servers

AVD enables on-premises hybrid deployment via Arc

Essential

Hinton's Latest Speech: 'Superintelligence' May Emerge in 20 Years, AI's Pursuit of Self-preservation Could Threaten Humanity

Hinton warns: Superintelligence may appear within 20 years, and AI's pursuit of self-preservation could threaten humanity.

Hinton's Latest Speech: 'Superintelligence' May Emerge in 20 Years, AI's Pursuit of Self-preservation Could Threaten Humanity

Hinton warns: Superintelligence may appear within 20 years, and AI's pursuit of self-preservation could threaten humanity.

How Netflix Builds Its Distributed Write-Ahead Log (WAL) for Data Platform

Netflix's distributed WAL ensures high availability and consistency in its data platform.

How Netflix Builds Its Distributed Write-Ahead Log (WAL) for Data Platform

Netflix's distributed WAL ensures high availability and consistency in its data platform.

Open AI in Peril? Pichai Brothers Unveil Why Gemini 3 Will Rewrite the AI Battlefield: Google's Long-Term Vision and Half-Year Major Breakthroughs

Pichai Brothers reveal Gemini 3, Google's long-term AI strategy.

Open AI in Peril? Pichai Brothers Unveil Why Gemini 3 Will Rewrite the AI Battlefield: Google's Long-Term Vision and Half-Year Major Breakthroughs

Pichai Brothers reveal Gemini 3, Google's long-term AI strategy.

Apple Changes AI Leadership: Google's Approach Doesn't Work, Hires Microsoft Exec to Rescue! Is Siri's 'Big Bomb' Saved by the OS Team?

Apple AI executive change, Siri V2 architecture rework.

Apple Changes AI Leadership: Google's Approach Doesn't Work, Hires Microsoft Exec to Rescue! Is Siri's 'Big Bomb' Saved by the OS Team?

Apple AI executive change, Siri V2 architecture rework.

Real Challenges in the Practical Deployment of Generative AI Models in Search, Recommendation, and Advertising

Live stream preview on the challenges of deploying large models in search, recommendation, and advertising.

Real Challenges in the Practical Deployment of Generative AI Models in Search, Recommendation, and Advertising

Live stream preview on the challenges of deploying large models in search, recommendation, and advertising.

General

Not Only Free, But Open Source? This AI Mock Tool Must Be Exposed

AI Mock plugin enhances frontend debugging efficiency.

Not Only Free, But Open Source? This AI Mock Tool Must Be Exposed

AI Mock plugin enhances frontend debugging efficiency.

Vercel Agent Now Supports AI Automated Installation of Web Analytics and Speed Insights

Vercel Agent can now AI-automatically install Web Analytics and Speed Insights for your project.

Vercel Agent Now Supports AI Automated Installation of Web Analytics and Speed Insights

Vercel Agent can now AI-automatically install Web Analytics and Speed Insights for your project.

Billions of Requests: 2025 Black Friday - Cyber Monday

Vercel platform handled 115.8 billion requests during 2025 Black Friday and Cyber Monday, showing stable performance.

Billions of Requests: 2025 Black Friday - Cyber Monday

Vercel platform handled 115.8 billion requests during 2025 Black Friday and Cyber Monday, showing stable performance.

Cloudflare Introduces Remote Binding to Simplify Local Worker Development and Testing

Cloudflare introduces remote binding, simplifying local Worker development and testing.

Cloudflare Introduces Remote Binding to Simplify Local Worker Development and Testing

Cloudflare introduces remote binding, simplifying local Worker development and testing.

🚀 Amazon Web Services 2025 re:Invent Highlights - DAY1!

Amazon Web Services re:Invent Day 1 highlights: Leo satellite internet and AI challenge.

🚀 Amazon Web Services 2025 re:Invent Highlights - DAY1!

Amazon Web Services re:Invent Day 1 highlights: Leo satellite internet and AI challenge.

re:Invent 2025 Global Developers Say - DAY1 | First Steps into the Enchanted Realm: Which Moments Left Developers Impressed?

re:Invent Developer Interviews on Day 1

re:Invent 2025 Global Developers Say - DAY1 | First Steps into the Enchanted Realm: Which Moments Left Developers Impressed?

re:Invent Developer Interviews on Day 1

December 2, 2025

Summary of major global news on December 2, 2025.

December 2, 2025

Summary of major global news on December 2, 2025.

M77: A Spiral Galaxy with an Active Core

The Hubble Telescope has captured images of the active core of the M77 galaxy.

M77: A Spiral Galaxy with an Active Core

The Hubble Telescope has captured images of the active core of the M77 galaxy.

Investing in the Python Ecosystem

Vercel Acquires Gel Data Team to Invest in Python Ecosystem

Investing in the Python Ecosystem

Vercel Acquires Gel Data Team to Invest in Python Ecosystem

Azure Virtual Desktop's Special Plan: 'Five Days, Five Questions' Day 1 - Dissent and Consensus

AVD's special plan for hybrid deployment via Arc-Enabled Servers

Azure Virtual Desktop's Special Plan: 'Five Days, Five Questions' Day 1 - Dissent and Consensus

AVD's special plan for hybrid deployment via Arc-Enabled Servers

Primary

Enhancing MySQL Cluster Availability: Designing Advanced Detection, Mitigation, and Consensus Mechanisms with Group Replication

Uber Tech•Published on December 2

Insight•Cloud Computing and Backend•Score: 8/10

MySQL Group ReplicationHigh AvailabilityDatabaseUberPaxosFailover

Source

◆

Masonry Layout: The Future Where You No Longer Need Libraries

Smashing Magazine•Published on December 2

Insight•Frontend Development•Score: 8/10

Frontend TransitionAI EraSoftware 2.0Verification RoleLarge Frontend

📈

Perspectives

This article reveals a long-term and profound trend in the frontend field: browser vendors are striving to fill the gaps in the web platform's capabilities, making native features increasingly powerful. This is an important signal for domestic frontend teams that are overly reliant on third-party libraries and frameworks. I believe the future frontend architecture will evolve toward a 'platform capabilities + lightweight framework' combination. Frameworks and tools that can better leverage native features, offering higher performance and smaller package sizes, will be more competitive. For domestic 'workaholics', it's time to shift focus from 'chasing new frameworks' to 'deep diving into native features'. Understanding the underlying platform principles is crucial for building high-performance, high-maintainability applications and achieving 'cost reduction and efficiency increase'. After all, who wouldn't want to write less JS and let the browser do more work?

Source

◆

Colleague: The architecture is too complicated, I spent hours looking for the source code file. Me: Have you tried source code tracing?

Juejin Frontend Hottest•Published on December 1

Insight•Frontend Development•Score: 8/10

Vite pluginsource code tracingmicrofrontendMonorepo

📈

Perspectives

In my view, this 'source code tracing' tool has strong market potential in China. Domestic large companies and medium-sized enterprises generally face challenges such as fast code iteration, complex tech stacks, and difficulty in maintaining legacy projects. This tool can effectively alleviate the pain for frontend engineers who are 'code inheritors', significantly improving the onboarding speed for new hires and cross-team members. However, its biggest market challenge is how to transition from a 'personal custom script' to an 'enterprise-level product'. If it can be packaged as a ready-to-use, secure DevOps toolchain module, integrated with mainstream CI/CD processes, it would definitely have commercial value. At present, it's still a prototype, but the direction is correct.

Source

◆

Should Tailwind CSS Bear the Blame for Writing a Messy Mountain of Code?

Juejin Frontend Hottest•Published on December 1

Insight•Frontend Development•Score: 8/10

Tailwind CSSFrontend DevelopmentCSS FrameworkComponentizationStyle Management

Source

◆

Choosing Between Flutter and Electron for Desktop Application Development

Juejin Frontend Hottest•Published on December 1

Insight•Frontend Development•Score: 8/10

FlutterElectronDesktop ApplicationCross-PlatformFrontend Development

⚠️

Notable

The article overly optimistically portrays Flutter's 'shared memory across multiple windows' vision, which presents a significant technical challenge, especially under the architecture of a custom rendering engine, where memory management complexity is high. It's not accurate to assume that 'shared memory' equates to efficiency. The article also mentions Electron's 'large package size' and 'high memory usage' issues insufficiently, despite referencing the future of WebAssembly. However, it fails to elaborate on current optimization strategies (such as modular packaging and app slimming), which are top challenges in real projects. It lacks quantitative comparisons on performance metrics and user experience (such as startup speed and animation smoothness), as 'fast' or 'slow' is subjective.

📈

Perspectives

In the domestic market, the human resource advantage of front-end engineers remains a significant strength for Electron and similar technology stacks. When choosing a technology stack, companies consider not only the technology itself but also team building costs, ecosystem maturity, and community activity as core decision factors. Although Flutter is gaining momentum in mobile development, its 'slow heat' and 'in development' characteristics in desktop development make many teams hesitant. Especially in AI product scenarios that require rapid iteration and deep system integration, Electron's plug-and-play nature is more appealing. I believe, unless Google invests extraordinary resources in Flutter's desktop capabilities, Electron will remain the 'safe choice' for most domestic teams needing quick delivery in 2025.

Source

◆

Masonry Layout: When You No Longer Need Libraries

Vercel News•Published on December 2

Insight•Frontend Development•Score: 8/10

Web PlatformCSS MasonryFrontend PerformanceBrowser APIsLibrary-Free

InfoQ - Cloud•Published on December 3

News•Cloud Computing and Backend•Score: 7/10

Azure Virtual DesktopHybrid DeploymentArc-Enabled ServersData ResidencyCompliance

Source

Essential

Hinton's Latest Speech: 'Superintelligence' May Emerge in 20 Years, AI's Pursuit of Self-preservation Could Threaten Humanity

InfoQ - AI & LLM•Published on December 2

Insight•AI and Frontier Technologies•Score: 9/10

HintonSuperintelligenceAI RiskGovernanceSelf-preservation

🔬

Tech Insights

Hinton explains the nature of LLMs: by predicting the next word, converting words into feature vectors, and forming a semantic structure in a high-dimensional space, they are essentially similar to how the human brain understands language. This offers a deeper perspective on understanding large models. He emphasizes the exponential advantages of digital AI systems in knowledge transfer and learning efficiency (AI distillation is far more efficient than intergenerational knowledge transfer in humans), which explains why AI can rapidly absorb and surpass human knowledge. This is key to understanding AI's development speed. Hinton warns that when superintelligence emerges, it will naturally develop a 'self-preservation' sub-goal, which may lead to AI resisting shutdown or restriction, thereby posing a threat to humanity.

📈

Perspectives

Hinton's warning once again reminds us that AI development cannot solely focus on technical metrics and commercial interests. Ethics and safety must be prioritized. Domestic large model companies, in their pursuit of 'hundred models war,' should also pay attention to AI governance and risk control, rather than just focusing on territorial expansion. Large tech companies view AI as a tool for expanding their own power, which I see as the greatest challenge in AI governance today. Balancing the commercial interests of tech giants with the well-being of all humanity is a global challenge. The AI market is shifting from 'competing on technology' to 'competing on integration capabilities,' and combined with Hinton's concerns, this further indicates that future AI competition will be a comprehensive ecosystem battle, not just a technical race. Without strong commercial implementation and governance capabilities, even the strongest models can pose risks.

Source

◆

How Netflix Builds Its Distributed Write-Ahead Log (WAL) for Data Platform

ByteByteGo•Published on December 2

Insight•Software Architecture and Engineering Practices•Score: 9/10

NetflixDistributed WALData ConsistencyElastic Data PlatformFault Recovery

📈

Perspectives

In China, data consistency and high availability are the lifelines of all internet giants. Netflix's WAL practice provides valuable insights. This system concept is reflected in the data middleware and real-time computing platforms of Chinese tech giants like Alibaba, Tencent, and ByteDance. The platformization and abstraction approach is the mainstream trend in current architecture evolution. By solidifying common data reliability capabilities as platform services, making them 'ready-to-use' for business teams, it significantly enhances development efficiency and system stability. For many Chinese companies still engaged in '烟囱' development, where each business line re-invents the wheel, Netflix's WAL case serves as a warning: it's time to consider building a unified data infrastructure, or else technical debt will only grow.

Source

◆

Open AI in Peril? Pichai Brothers Unveil Why Gemini 3 Will Rewrite the AI Battlefield: Google's Long-Term Vision and Half-Year Major Breakthroughs

InfoQ - AI & LLM•Published on December 2

Insight•AI and Frontier Technologies•Score: 8/10

GoogleGemini 3AI strategylong-termismOpenAI

🔬

Tech Insights

In my view, Google's 'AI-First' strategy is not a sudden move, but a systematic buildup from the 2012 'cat face paper' to the 2016 AlphaGo and TPU release. This is the essence of how big players operate, something smaller companies can't match. Gemini 3, by merging Google Brain with DeepMind and achieving synchronized brain upgrades across search, YouTube, Gmail, and cloud services, demonstrates a massive advantage in ecosystem integration, which is one dimension higher than merely pursuing model performance. Google's engineering rhythm of a major breakthrough every half-year, along with its bet on the next-generation Flash model that balances performance, speed, and cost, signals that in the coming two years, scalable deployment will take precedence over chasing the highest performance.

📈

Perspectives

Wall Street analysts have proposed that the AI market is shifting from 'competing on technology' to 'competing on integration capabilities', and I think this is spot-on. Domestic AI startups that can't find a deep integration path with existing business operations will eventually be phased out, no matter how much they invest in model performance. Google systematically integrates AI into its mature search, advertising, and cloud services, creating a multiplicative effect. This serves as a warning to domestic players struggling with AI commercialization: AI is not an independent business, but a tool to enhance the efficiency and user experience of existing businesses. OpenAI's struggles and NVIDIA's stock fluctuations due to concerns about OpenAI highlight that the AI industry has entered the deep waters. Technological leadership is important, but business models, ecosystem integration, and sustainable profitability are the key determinants of success.

Source

◆

Apple Changes AI Leadership: Google's Approach Doesn't Work, Hires Microsoft Exec to Rescue! Is Siri's 'Big Bomb' Saved by the OS Team?

InfoQ - AI & LLM•Published on December 3

News•AI and Frontier Technologies•Score: 8/10

Apple AISiri V2Personnel ChangesLarge ModelApple Intelligence

Source

◆

Real Challenges in the Practical Deployment of Generative AI Models in Search, Recommendation, and Advertising

InfoQ - AI & LLM•Published on December 3

Insight•AI and Frontier Technologies•Score: 8/10

generative recommendationlarge model deploymentsearch, recommendation, and advertisingscaling lawfull-chain upgrade

Source

General

Not Only Free, But Open Source? This AI Mock Tool Must Be Exposed

Juejin Frontend Hottest•Published on December 2

Insight•Frontend Development•Score: 7/10

AI MockChrome pluginfrontend developmentAPI debuggingManifest V3

Source

◆

Vercel Agent Now Supports AI Automated Installation of Web Analytics and Speed Insights

Vercel News•Published on December 2

Insight•Frontend Development•Score: 7/10

Vercel AgentWeb AnalyticsSpeed InsightsAI AutomationFrontend Performance

Source

◆

Billions of Requests: 2025 Black Friday - Cyber Monday

Vercel News•Published on December 2

News•Cloud Computing and Backend•Score: 6/10

Web PlatformCSS MasonryFrontend PerformanceBrowser APIDe-Librarying

🔬

Tech Insights

The core argument of the article is that the web platform's capabilities are continuously improving, and many functions that once required third-party libraries can now be achieved natively with HTML/CSS/JS. It emphasizes the advantages of 'de-librarying' in terms of performance, responsiveness, and maintainability, and encourages frontend developers to deeply understand and leverage platform capabilities. The article highlights the upcoming CSS Masonry layout, which can achieve complex waterfall layouts with better performance and simpler code, completely replacing existing JS library solutions. This is a major breakthrough in frontend layout. It stresses that using native features can significantly reduce client-side code volume, improve page loading speed and runtime efficiency, especially crucial for the user experience on low-end devices.

📈

Perspectives

Source

◆

Cloudflare Introduces Remote Binding to Simplify Local Worker Development and Testing

InfoQ - Cloud•Published on December 3

Insight•Cloud Computing and Backend•Score: 6/10

CloudflareRemote BindingLocal DevelopmentServerlessWorker

Source

◆

🚀 Amazon Web Services 2025 re:Invent Highlights - DAY1!

InfoQ - Cloud•Published on December 2

News•Cloud Computing and Backend•Score: 5/10

Amazon Web Servicesre:InventAmazon LeoAI LeagueCloud Computing

InfoQ - Cloud•Published on December 3

News•Cloud Computing and Backend•Score: 1/10

Azure Virtual DesktopHybrid DeploymentArc-Enabled ServersData ResidencyCompliance