2025-11-19 Tech Briefing | AWS launches a zero-overage pricing plan for CloudFront.

Wednesday

Welcome to this edition, featuring 29 articles.

📚 Contents/📝 TL;DR

📚 Contents

📝 TL;DR

Primary

AWS Introduces a Zero-Overage Pricing Plan for CloudFront

AWS launches a zero-overage pricing plan for CloudFront.

AWS Introduces a Zero-Overage Pricing Plan for CloudFront

AWS launches a zero-overage pricing plan for CloudFront.

New Amazon Bedrock Service Tiers Help Balance AI Workload Performance and Cost

Bedrock introduces three service tiers to optimize AI cost and performance

New Amazon Bedrock Service Tiers Help Balance AI Workload Performance and Cost

Bedrock introduces three service tiers to optimize AI cost and performance

Windows to Become 'Agent OS'? Developers Strongly Oppose, CEO Closes Comments

Microsoft's push for Agent OS sparks developer backlash, with top executives responding to the out-of-control situation.

Windows to Become 'Agent OS'? Developers Strongly Oppose, CEO Closes Comments

Microsoft's push for Agent OS sparks developer backlash, with top executives responding to the out-of-control situation.

Building a Priority-Based Message Processing System Using Amazon MQ and AWS App Runner

AWS demonstrates how to build a priority-based message system using MQ and App Runner.

Building a Priority-Based Message Processing System Using Amazon MQ and AWS App Runner

AWS demonstrates how to build a priority-based message system using MQ and App Runner.

Cloudflare Proposes Using Merkle Tree Certificates to Address TLS Performance in the Post-Quantum Era

A proposal for optimizing TLS performance in the post-quantum era using Merkle tree certificates.

Cloudflare Proposes Using Merkle Tree Certificates to Address TLS Performance in the Post-Quantum Era

A proposal for optimizing TLS performance in the post-quantum era using Merkle tree certificates.

Accelerate Infrastructure Development with Pre-Deployment Validation and Simplified Troubleshooting via CloudFormation

CloudFormation introduces pre-validation and troubleshooting capabilities.

Accelerate Infrastructure Development with Pre-Deployment Validation and Simplified Troubleshooting via CloudFormation

CloudFormation introduces pre-validation and troubleshooting capabilities.

Dialogue with 'Zebra Oral Language': How the World's First AI Tutor Outperforms Humans?

Zebra Oral Language has launched an AI tutor based on its own developed model, offering an experience that surpasses that of human teachers.

Dialogue with 'Zebra Oral Language': How the World's First AI Tutor Outperforms Humans?

Zebra Oral Language has launched an AI tutor based on its own developed model, offering an experience that surpasses that of human teachers.

AWS Lambda Supports Python 3.14 Runtime

AWS Lambda now supports Python 3.14.

AWS Lambda Supports Python 3.14 Runtime

AWS Lambda now supports Python 3.14.

Modelfactory 020 Week AI Application Chart: Lybic Leads the Chart, Highlighting the 'Agent Native Infrastructure' Trend

Lybic leads the AI application chart, focusing on Agent native tool construction.

Modelfactory 020 Week AI Application Chart: Lybic Leads the Chart, Highlighting the 'Agent Native Infrastructure' Trend

Lybic leads the AI application chart, focusing on Agent native tool construction.

278 Billion Market Opportunity, MongoDB Game Data New Infrastructure Technology Unveiled! | Q Recommendation

Tencent Cloud will reveal MongoDB optimization technologies for game scenarios.

278 Billion Market Opportunity, MongoDB Game Data New Infrastructure Technology Unveiled! | Q Recommendation

Tencent Cloud will reveal MongoDB optimization technologies for game scenarios.

Big News! Kiro Upgrades Significantly, Developers Are Taking Notice!

Kiro Releases a Major Version Update

Big News! Kiro Upgrades Significantly, Developers Are Taking Notice!

Kiro Releases a Major Version Update

Musk Outpaces Google with Grok 4.1, Tops AI Authority Rankings, and We Tested It Too

Grok 4.1 tops AI rankings, Musk leads again.

Musk Outpaces Google with Grok 4.1, Tops AI Authority Rankings, and We Tested It Too

Grok 4.1 tops AI rankings, Musk leads again.

OceanBase Open Sources Its First AI Database SeekDB: Follows Apache 2.0 License, Enables AI Application Development in Three Lines of Code

OceanBase opens sources its first AI database SeekDB, supporting AI application development in three lines of code.

OceanBase Open Sources Its First AI Database SeekDB: Follows Apache 2.0 License, Enables AI Application Development in Three Lines of Code

OceanBase opens sources its first AI database SeekDB, supporting AI application development in three lines of code.

Essential

How Disney Hotstar (Now JioHotstar) Scales Its Infrastructure for 60 Million Concurrency Users

Hotstar achieved scalability for 60 million concurrent users through architecture refactoring.

How Disney Hotstar (Now JioHotstar) Scales Its Infrastructure for 60 Million Concurrency Users

Hotstar achieved scalability for 60 million concurrent users through architecture refactoring.

Refactoring a Class, JVM Saved 2.9GB of Memory?

Refactoring data structures reduced JVM memory from 3.13GB to 211MB.

Refactoring a Class, JVM Saved 2.9GB of Memory?

Refactoring data structures reduced JVM memory from 3.13GB to 211MB.

Cloudflare's Major Outage on November 18, 2025

Cloudflare experienced a global outage for nearly 6 hours due to a database permission change.

Cloudflare's Major Outage on November 18, 2025

Cloudflare experienced a global outage for nearly 6 hours due to a database permission change.

New Amazon EC2 P6-B300 Instance Accelerates Large-Scale AI Applications

AWS launches P6-B300 instance with NVIDIA Blackwell GPU

New Amazon EC2 P6-B300 Instance Accelerates Large-Scale AI Applications

AWS launches P6-B300 instance with NVIDIA Blackwell GPU

CodeClash: Benchmarking Large Language Models through Multi-Round Programming Competitions

CodeClash introduces a new benchmark for evaluating LLMs' advanced programming capabilities.

CodeClash: Benchmarking Large Language Models through Multi-Round Programming Competitions

CodeClash introduces a new benchmark for evaluating LLMs' advanced programming capabilities.

Stop Fixating on Frameworks! Your Technical Roadmap Needs an Update

Frontend needs to shift from frameworks to full-stack capabilities.

Stop Fixating on Frameworks! Your Technical Roadmap Needs an Update

Frontend needs to shift from frameworks to full-stack capabilities.

From 'Ambient Programming' to 'Ambient Engineering': Kiro Turns Every Developer into a '100-Person Team'

Kiro Promotes a New Paradigm for AI-Assisted Development

From 'Ambient Programming' to 'Ambient Engineering': Kiro Turns Every Developer into a '100-Person Team'

Kiro Promotes a New Paradigm for AI-Assisted Development

November 18, 2025: Global Tech and Political Dynamics Summary

Summary of global tech and political dynamics.

November 18, 2025: Global Tech and Political Dynamics Summary

Summary of global tech and political dynamics.

After Interviewing Others, I've Lost My Faith in Interviews

A frontend engineer shares observations and reflections on interviews.

After Interviewing Others, I've Lost My Faith in Interviews

A frontend engineer shares observations and reflections on interviews.

General

Kgateway v2.1 Released!

Kgateway 2.1 is released, integrating Agentgateway support for AI workloads.

Kgateway v2.1 Released!

Kgateway 2.1 is released, integrating Agentgateway support for AI workloads.

Analyzing AWS Network Firewall Logs with Amazon OpenSearch Dashboard

Simplify AWS firewall log analysis and enhance security visualization

Analyzing AWS Network Firewall Logs with Amazon OpenSearch Dashboard

Simplify AWS firewall log analysis and enhance security visualization

After Interviewing Others, I've Lost My Enchantment with Interviews

A frontend engineer shares observations and reflections on interviews.

After Interviewing Others, I've Lost My Enchantment with Interviews

A frontend engineer shares observations and reflections on interviews.

Amazon ECS New Features at AWS re:Invent 2025

AWS previews multiple AI integration and deployment optimization features for ECS at re:Invent 2025.

Amazon ECS New Features at AWS re:Invent 2025

AWS previews multiple AI integration and deployment optimization features for ECS at re:Invent 2025.

Preface of 'Frictionless': Redefining Productivity

Preface for a new book on R&D efficiency.

Preface of 'Frictionless': Redefining Productivity

Preface for a new book on R&D efficiency.

Answering Developers' Questions: HarmonyOS Technical Analysis, Part 2

HarmonyOS Developer Q&A, no original content available

Answering Developers' Questions: HarmonyOS Technical Analysis, Part 2

HarmonyOS Developer Q&A, no original content available

“Answering Developers’ Questions” on HarmonyOS Technical Issues - Issue 3

No original content available for the HarmonyOS developer Q&A section.

“Answering Developers’ Questions” on HarmonyOS Technical Issues - Issue 3

No original content available for the HarmonyOS developer Q&A section.

Primary

AWS Introduces a Zero-Overage Pricing Plan for CloudFront

AWS 网络•Published on November 18

Insight•Cloud Computing and Backend•Score: 9/10

CloudFrontFixed-Rate PricingDDoS ProtectionCost ForecastingPay-as-You-Go

Source

◆

New Amazon Bedrock Service Tiers Help Balance AI Workload Performance and Cost

AWS 新闻•Published on November 18

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

Amazon BedrockService TiersAI WorkloadsCost OptimizationLLM Inference

Source

◆

Windows to Become 'Agent OS'? Developers Strongly Oppose, CEO Closes Comments

InfoQ - AI & LLM•Published on November 18

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

Windows AIizationAgent OSDeveloper ResistanceMicrosoft StrategyOperating System Evolution

Source

◆

Building a Priority-Based Message Processing System Using Amazon MQ and AWS App Runner

AWS - Architecture•Published on November 18

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

AWSAmazon MQApp RunnerMessage QueuePriority Processing

Source

◆

Cloudflare Proposes Using Merkle Tree Certificates to Address TLS Performance in the Post-Quantum Era

InfoQ - Cloud•Published on November 18

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

Merkle TreeTLSPost-Quantum CryptographyPerformance Optimization

Source

◆

Accelerate Infrastructure Development with Pre-Deployment Validation and Simplified Troubleshooting via CloudFormation

AWS DevOps•Published on November 18

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

CloudFormationIaCPre-Deployment ValidationTroubleshootingDevOps

Source

◆

Dialogue with 'Zebra Oral Language': How the World's First AI Tutor Outperforms Humans?

Book Flips•Published on November 18

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

AI tutorZebra Oral Languagechild educationproprietary modelemotion recognition

Source

◆

AWS Lambda Supports Python 3.14 Runtime

AWS 计算•Published on November 18

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

AWS LambdaPython 3.14ServerlessContainer SupportPerformance Optimization

Source

◆

Modelfactory 020 Week AI Application Chart: Lybic Leads the Chart, Highlighting the 'Agent Native Infrastructure' Trend

InfoQ - AI & LLM•Published on November 18

News•AI and Frontier Technologies•Score: 7/10

AI application chartAgent nativeLybicAI workflowintelligent infrastructure

Source

◆

278 Billion Market Opportunity, MongoDB Game Data New Infrastructure Technology Unveiled! | Q Recommendation

InfoQ - Big Data•Published on November 19

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

MongoDBgame databasekey rollbacklossless migrationhigh concurrency rate limiting

Source

◆

Big News! Kiro Upgrades Significantly, Developers Are Taking Notice!

InfoQ - AI & LLM•Published on November 18

News•AI and Frontier Technologies•Score: 6/10

KiroAI Development ToolsFeature UpgradeDeveloper Ecosystem

Source

◆

Musk Outpaces Google with Grok 4.1, Tops AI Authority Rankings, and We Tested It Too

InfoQ - AI & LLM•Published on November 18

News•AI and Frontier Technologies•Score: 1/10

GrokAILarge ModelsMuskRankings

Source

◆

OceanBase Open Sources Its First AI Database SeekDB: Follows Apache 2.0 License, Enables AI Application Development in Three Lines of Code

InfoQ - Backend•Published on November 18

News•AI and Frontier Technologies•Score: 1/10

OceanBaseAI DatabaseSeekDBOpen SourceApache 2.0

📈

Perspectives

In November 2025, the domestic database market is highly competitive, ranging from traditional relational databases to cloud-native databases, and emerging vector databases. OceanBase, as an important product within the Alibaba ecosystem, chose to open-source its AI database SeekDB at this time, aiming to seize the AI + database new market. The open-sourcing under the Apache 2.0 license also shows its openness to the community. I believe this approach of embedding AI capabilities into the database layer may become a key direction for future database development. It can lower the threshold for enterprises to build AI applications, especially for small and medium-sized businesses lacking dedicated AI teams. However, the market acceptance will depend on its actual performance, ecosystem support, and usability.

Source

Essential

How Disney Hotstar (Now JioHotstar) Scales Its Infrastructure for 60 Million Concurrency Users

ByteByteGo•Published on November 18

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

High ConcurrencyKubernetesEKSCDNMicroservicesData Center AbstractionEnvoy

🔬

Tech Insights

1. Deep Optimization of CDN as API Gateway: Not just caching, but also handling security, rate limiting, and routing. By separating cacheable and non-cacheable APIs, adjusting refresh rates, and simplifying rules, the CDN's computational power is maximized, effectively relieving backend pressure. 2. Precision Infrastructure Expansion: Addressing the traffic bottleneck of NAT Gateway by shifting from 'one per AZ' to 'one per subnet', and Kubernetes worker nodes using 'topology spread constraints' to avoid network hotspots, demonstrates meticulous control over details. 3. Data Center Abstraction and Envoy Central Proxy: This is the most impressive part. By logically abstracting multiple K8s clusters into a single 'data center', combined with Envoy to build a central proxy layer, it unifies routing, authentication, and rate limiting, completely resolving the 'traditional pain points' of NodePort exhaustion, IP address shortages, load balancer preheating, and achieving true cluster seamless deployment.

📈

Perspectives

In my view, Hotstar's case is an excellent textbook for the 'high concurrency architecture practice' of domestic internet giants. Especially at the time point of November 2025, the architecture of leading players in live streaming, short video, and e-commerce fields in China has largely evolved to a similar level, and even surpassed it. The 'data center abstraction' concept essentially represents the extreme embodiment of platform engineering thinking, aiming to provide application teams with a simpler and more unified deployment and operation experience. This approach also verifies the core role of managed Kubernetes and high-performance API Gateways (such as Envoy) in the cloud-native era. For teams still struggling with self-built Kubernetes, this is a clear slap: embrace managed services and focus your efforts on business innovation rather than fighting with infrastructure complexity.

Source

◆

Refactoring a Class, JVM Saved 2.9GB of Memory?

Alibaba Cloud Tech•Published on November 12

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

JVM Memory OptimizationFastUtilLong2ObjectOpenHashMapData Structure SelectionObject Overhead

Source

◆

Cloudflare's Major Outage on November 18, 2025

Cloudflare Blog•Published on November 18

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

CloudflareService OutageConfiguration ErrorClickHouseBot Management

Source

◆

New Amazon EC2 P6-B300 Instance Accelerates Large-Scale AI Applications

AWS 新闻•Published on November 18

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

Amazon EC2P6-B300NVIDIA Blackwell UltraLarge Model TrainingEFAGPU Computing Power

Source

◆

CodeClash: Benchmarking Large Language Models through Multi-Round Programming Competitions

Alibaba Cloud Tech•Published on November 16

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

LLM BenchmarkingCodeClashProgramming AbilityCompetitive ProgrammingModel Evaluation

🔬

Tech Insights

CodeClash's core lies in simulating 'high-level goals'-driven software development, which better reflects LLMs' engineering capabilities in complex systems than traditional unit testing or bug fixing. I believe this is the truly valuable direction for evaluation, as we ultimately seek AI that can solve real-world problems, not just write demos. It introduces iterative development and competitive mechanisms, allowing LLMs to optimize code through multiple rounds of feedback, closely mirroring real-world development workflows. This 'fight and practice' model helps us observe how models perform in continuous learning and adaptability. Research shows that even top models like GPT 5 and Claude Sonnet 4.5 fail to demonstrate absolute superiority in one-to-many competitions, indicating significant room for improvement in complex collaboration and strategic competition.

Source

◆

Stop Fixating on Frameworks! Your Technical Roadmap Needs an Update

Juejin Frontend Hottest•Published on November 18

Insight•Frontend Development•Score: 9/10

Frontend EvolutionTechnical RoadmapAI EmpowermentEngineering CapabilitiesCross-Platform Development

Source

◆

From 'Ambient Programming' to 'Ambient Engineering': Kiro Turns Every Developer into a '100-Person Team'

InfoQ - AI & LLM•Published on November 18

Insight•AI and Frontier Technologies•Score: 7/10

KiroAmbient ProgrammingAI-Assisted DevelopmentProductivity Tools

Source

◆

November 18, 2025: Global Tech and Political Dynamics Summary

Global Vision•Published on November 18

News•Comprehensive News•Score: 6/10

Satellite LaunchRocket DelayEconomic DataInternational RelationsMilitary Arms Trade

Source

◆

After Interviewing Others, I've Lost My Faith in Interviews

InfoQ - AI & LLM•Published on November 18

News•Software Architecture & Engineering Practices•Score: 1/10

Interview ExperienceCommunication ExpressionTechnical EvaluationEducational BiasFrontend Hiring

Source

General

Kgateway v2.1 Released!

CNCF•Published on November 18

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

Kubernetes Gateway APIKgatewayAgentgatewayAI GatewayCNCFEnvoy

🔬

Tech Insights

1. Innovative Integration of AI Gateway: Through Agentgateway, Kgateway v2.1 has achieved the first implementation of AI connectivity capabilities for LLMs, MCP tools, and AI agents. This indicates that the Kubernetes ecosystem is actively embracing AI workloads, integrating AI capabilities into the infrastructure layer, signaling a "future trend". 2. Continuous Evolution of Gateway API: Kgateway is fully compatible with Kubernetes Gateway API 1.3.0 and Inference Extension 1.0.0, further solidifying the Gateway API's position as an "ingress alternative" and the next-generation standard for traffic management. 3. Enterprise Feature Enhancements: Global policy attachment, deep merging and priority control of ExtAuth/ExtProc policies, HPA plugin support, dynamic forwarding proxy, session affinity, enhanced retry and timeout mechanisms, and passive health checks all demonstrate that Kgateway is moving towards "production readiness" and enterprise-grade features.

📈

Perspectives

In my view, the release of Kgateway v2.1, especially its integration of "AI Gateway", is a notable development in the cloud-native space this year (2025). This signals that API Gateways are no longer just traditional traffic entry points, but are evolving to support more "intelligent routing" and "AI service governance" capabilities. Domestic cloud providers and leading enterprises are also accelerating their layout in AI infrastructure. Kgateway's step in this direction undoubtedly provides the industry with a new perspective on how to more naturally integrate AI capabilities into existing cloud-native traffic management systems. However, it's important to remember that any new feature in Beta may carry risks, and the cost of "blindly chasing new features" can be significant. A prudent architect would first observe and then act.

Source

◆

Analyzing AWS Network Firewall Logs with Amazon OpenSearch Dashboard

AWS - Security•Published on November 18

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

AWS Network FirewallOpenSearchLog AnalysisSecurity MonitoringCloudWatch

Source

◆

After Interviewing Others, I've Lost My Enchantment with Interviews

Juejin Frontend Hottest•Published on November 18

Insight•Software Architecture and Engineering Practices•Score: 6/10

Interview ExperienceCommunication SkillsTechnical EvaluationEducational BiasFrontend Recruitment

Source

◆

Amazon ECS New Features at AWS re:Invent 2025

AWS - Containers•Published on November 18

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

Amazon ECSre:Invent 2025AI DevelopmentMCP ProtocolBlue-Green Deployment

Source

◆

Preface of 'Frictionless': Redefining Productivity

Martin Fowler•Published on November 18

News•R&D Efficiency•Score: 4/10

ProductivityDevExNicole Forsgren

Source

◆

Answering Developers' Questions: HarmonyOS Technical Analysis, Part 2

InfoQ - Frontend•Published on November 18

Insight•Software Architecture and Engineering Practice•Score: 1/10

HarmonyOSDeveloper Q&ATechnical Analysis

Source

◆

“Answering Developers’ Questions” on HarmonyOS Technical Issues - Issue 3

InfoQ - Frontend•Published on November 18

Insight•Software Architecture and Engineering Practice•Score: 1/10

HarmonyOSdeveloper questionstechnical analysis