ByteByteGo Source Map

Vai trò

Note này là điểm nối giữa nguồn thô ByteByteGo và knowledge graph của vault. Source gốc được giữ ở:

• 00 - Sources/ByteByteGo/Articles
• 00 - Sources/ByteByteGo/Images

Không biến toàn bộ source thành summary dài. Chỉ rút ra concept/synthesis khi bài có định nghĩa, cơ chế, trade-off hoặc pattern có thể tái dùng.

Lớp concept đã tạo

• API Gateway
• Rate Limiting
• Throttling
• DNS
• Reverse Proxy
• Web Request Path
• TLS Termination
• Health Check
• Cache Stampede
• Edge Function
• Idempotency Key
• Load Balancer
• High Availability
• Observability
• Data Replication
• Database Sharding
• Message Broker
• Message Queue
• Publish-Subscribe
• Event Stream
• Apache Kafka
• Kafka Partition
• Consumer Group
• Delivery Semantics
• Dead Letter Queue
• RabbitMQ
• Apache Pulsar
• Microservices Design Patterns
• Docker
• Containerization
• Virtualization
• Container Image
• Container Runtime
• Linux Namespace
• Control Groups
• Serverless Architecture
• Function as a Service
• Backend as a Service
• Serverless Cold Start
• Lambda Execution Environment
• Firecracker MicroVM
• Lambda SnapStart
• Provisioned Concurrency
• Serverless Cost Model
• Lambda Layer
• Serverless Worker Sharding
• Monolithic Architecture
• Modular Monolith
• Microservices Architecture
• Sidecar Pattern
• Ambassador Pattern
• Container Adapter Pattern
• Work Queue Pattern
• Scatter-Gather Pattern
• Eventual Consistency
• Caching Strategy
• Redis
• Distributed Cache
• Redis Data Structures
• Redis Event Loop
• Redis Persistence
• Cache-Aside
• Read-Through Cache
• Write-Through Cache
• Write-Behind Cache
• Cache Invalidation
• Cache Eviction Policy
• Redis Streams
• Redis Sorted Set
• HyperLogLog
• Distributed Lock
• Cache Warmup
• Feature Store Cache
• ML Platform
• Internal Platform as Product
• Feature Store
• Offline Feature Store
• Online Feature Store
• Training-Serving Skew
• Feature Discovery
• Data Catalog
• Model Self-Test
• Regression Testing
• Model Shadowing
• Prediction Logging
• Model Feedback Loop
• Incremental Model Training
• Feature Drift
• Feature Collocation
• Prediction Serving Fanout
• Inference Compute Graph Split
• Raw Feature Transport
• Model Deployment Reconciliation
• Model Onboarding
• Managed Model Serving Integration
• Vendor Lock-In
• Annotation Platform
• Human-in-the-Loop Labeling
• Annotation Quality Metrics
• Annotation Debt
• Content Delivery Network
• REST API
• GraphQL
• GraphQL Federation
• Database Indexing
• Database Schema Design
• Retry Pattern
• Service Mesh
• CAP and PACELC
• Backpressure
• Circuit Breaker
• Consistent Hashing
• Distributed Systems
• Horizontal Scaling
• Service Discovery
• Retrieval-Augmented Generation
• Agentic RAG
• LLM Agent
• Agentic Loop
• Tool Use
• Function Calling
• Model Context Protocol
• Context Engineering
• LLM Memory
• KV Cache
• LLM Inference Engineering
• LLM Evaluation
• LLM Observability
• Agent Evaluation
• Agent Tracing
• LLM-as-Judge
• Evidence-Grounded Generation
• Retrieval Evaluation
• Hybrid Retrieval
• Citation Quality
• Multi-Agent System
• AI Hallucination
• LLM Security
• Prompt Injection
• Excessive Agency
• Preference Learning
• Multimodal LLM
• AI Hardware Accelerator
• AI Model Serving
• Agent Harness
• Coding Agent
• Agent Orchestrator-Worker Pattern
• Citation Agent
• Agent Effort Scaling
• Agent State Checkpointing
• Diff Problem
• Sandboxed Agent Execution
• Zero-Secret Agent Architecture
• Safe Outputs Pipeline
• Agent Trust Boundary Logging
• Agent Protocol Interoperability
• Agent Evaluation Stack
• Graph RAG
• Model Router
• AI Search
• Feed Retrieval
• Semantic Retrieval
• Generative Retrieval
• Recommendation Funnel
• Cold Start Problem
• Ranking
• Query Understanding
• Two-Tower Retrieval
• Vector Search Infrastructure
• Commonsense Knowledge Graph
• Product Recommendation System
• Multimodal Search
• Foundation Model for Recommendation
• Multimodal Annotation Fusion
• Embedding Lifecycle Management
• LLM Architecture Comparison
• Structured Logging
• Metrics
• Metric Cardinality
• Distributed Tracing
• Service Level Indicator
• Service Level Objective
• Error Budget
• Capacity Planning
• Peak QPS
• Load Testing
• Control Plane
• Data Plane
• Hidden Dependency
• Herd Effect
• Kill Switch
• Fail Closed
• Cold Read Path
• Replication Is Not Backup
• Status Page Dependency
• Global Metadata Replication
• Continuous Integration
• Continuous Delivery
• Continuous Deployment
• Deployment Pipeline
• Big-Bang Deployment
• Rolling Deployment
• Blue-Green Deployment
• Canary Deployment
• Feature Flag
• Dark Launch
• Shadow Traffic
• Expand-Contract Migration
• Rollback Alarm
• Bake Period
• Phased Rollout
• Incident Response
• Postmortem
• Premortem
• Root Cause Analysis
• Risk Matrix
• Rollback Strategy
• Disaster Recovery
• Backup and Restore
• Alerting
• API Contract
• API Lifecycle Management
• API Documentation
• API Composition
• API Aggregation
• API Orchestration
• Backend for Frontend
• API Protocol
• gRPC
• SOAP
• Async API Pattern
• Client State Synchronization
• Streaming Compression
• Delta Update
• Passive Session
• Mobile Bandwidth Optimization
• Short Polling
• Long Polling
• Server-Sent Events
• WebSocket
• Webhook
• GraphQL Subscription
• API Versioning
• API Pagination
• API Security
• Stateless Architecture
• Payment Intent
• Payment Method
• Payment State Machine
• Local Payment Method
• Payment Orchestration
• Payment Service Provider
• Fraud Detection System
• Precision-Recall Tradeoff
• Backward Compatibility
• Authentication
• Authorization
• Fine-Grained Authorization
• Attribute-Based Access Control
• Relationship-Based Access Control
• Google Zanzibar
• Permission Tuple
• Authorization Consistency Token
• Policy Information Point
• Token Exchange
• Federated Identity Provider
• Least Privilege
• JSON Web Token
• Session-Based Authentication
• PASETO
• OAuth 2.0
• OpenID Connect
• Single Sign-On
• Multi-Factor Authentication
• Input Validation
• SQL Injection
• Cross-Site Scripting
• Cross-Site Request Forgery
• Database Transaction
• ACID
• Transaction Isolation
• Search Engine Architecture
• Inverted Index
• Index Segment
• Search Indexer
• Search Broker
• Search Tenant Isolation
• Cell-Based Architecture
• Destination-Aware Batching
• Search Query AST
• Search Ranking
• Zero-Downtime Reindexing
• Query Execution Plan
• Query Planner
• Full Table Scan
• Database Partitioning
• SQL Database
• NoSQL Database
• NewSQL
• Document Store
• Join Operation
• Relational Database Design
• MVCC
• Snapshot Isolation
• Deadlock
• Database Workload Isolation
• Read Path
• Write Path
• Staleness
• Read-Your-Writes Consistency
• Read Replica
• Materialized View
• Specialized Read Store
• Change Data Capture
• Transactional Outbox
• CQRS
• Fan-Out on Write
• Fan-Out on Read
• Strong Consistency
• Linearizability
• Serializability
• Strict Serializability
• Consensus
• Quorum
• Leader Election
• Raft
• Paxos
• Saga Pattern
• Concurrency Control
• Pessimistic Locking
• Optimistic Locking
• Write-Ahead Log
• Storage Engine
• B-Tree
• LSM Tree
• Compaction
• Bloom Filter
• Write Amplification
• Read Amplification
• Space Amplification
• Cassandra
• Distributed Counter
• Rollup Pipeline
• Event Log
• Time-Series Data Storage
• Data Lifecycle Management
• Financial Source of Truth
• Data Contract
• Shadow Testing
• Data Freshness
• Batch Processing
• Micro-Batch Processing
• Stream Processing
• Event Time and Processing Time
• Data Processing Window
• Watermark
• Late Data
• Lambda Architecture
• Kappa Architecture
• Snapshot Bootstrap
• Data Pipeline Validation
• Spark on Kubernetes Platform
• Remote Shuffle Service
• Data Platform as Code
• Data Warehouse
• Data Lake
• Data Mesh
• Object Storage
• Amazon S3
• Object Metadata Index
• Storage Class Tiering
• Multipart Upload
• Distributed Key-Value Store
• Derived Data Store
• Cost Optimization
• Unified Domain Model
• Real-Time Graph Architecture
• Property Graph
• Key-Value Graph Storage
• Retry Storm
• Timeout
• Latency
• Failover
• Partial Failure
• Gray Failure
• Cascading Failure
• Load Shedding
• Bulkhead Pattern
• Metastable Failure
• Correlated Failure
• Graceful Degradation
• Blast Radius
• Synthetic Monitoring
• Chaos Engineering
• Debugging as Code
• Runbook Automation
• Automated Root Cause Analysis
• Analyzer Chaining
• Dependency Graph
• Context Gathering
• Diagnostic Agent
• Production State Replay
• Indirect Prompt Injection
• Lethal Trifecta
• Agents Rule of Two
• Untrusted Model Output
• LLM Supply Chain Security
• RAG Knowledge Base Poisoning
• Video Streaming Architecture
• Adaptive Bitrate Streaming
• Video Transcoding Pipeline
• Proactive Caching
• Live Streaming Origin
• In-Memory Read Model
• Workflow Orchestration
• Kubernetes
• Declarative Reconciliation
• Kubernetes Pod
• Kubernetes Controller
• Kubernetes Service
• Kubernetes Operator Pattern
• Kubernetes Autoscaling
• Stateful Workloads on Kubernetes
• Kubernetes Load Balancing
• Zero-Downtime Infrastructure Migration
• Infrastructure as Code
• Java Virtual Threads
• Generational Garbage Collection
• Runtime Platform Migration
• Technical Debt
• Legacy System Modernization
• Traffic Replay
• State Reconciliation Pipeline
• Behavioral Compatibility
• Codemod Migration
• Dependency-Driven Migration
• Leaf-to-Root Migration
• Service Layer Refactoring
• Intent-Based Test Migration
• Idiomatic Rewrite
• Dual-System Operation
• Critical Path Build Graph
• Mobile App Modularization
• Compiler-Driven Modularization
• Developer Velocity
• Hostile Multi-Tenancy
• Multi-Tenancy
• Tenancy Isolation Spectrum
• Tenant Storage Model
• Tenant Context
• Noisy Neighbor Problem
• Cross-Tenant Data Leak
• Resource Quota
• Sandboxed Build Execution
• Build Provisioning Warm Pool
• Notification Budgeting
• Notification Recommender Pipeline
• Reranking
• Causal Inference
• A-B Testing
• Design-to-Code Context
• Code Connect
• Component Mapping
• Design System Drift
• Event Sourcing
• JVM Architecture
• Direct Preference Optimization
• Model Alignment
• Model Distillation
• Logical Clocks
• Lamport Timestamps
• Vector Clocks
• TrueTime
• Proof of Personhood
• Sybil Resistance
• World Models
• Reasoning Model
• Agent Communication Protocol
• A2A Protocol
• Voice AI Infrastructure
• Clickbait Filtering
• Bot Management
• AI-Native Developer Platform
• Context Compression
• LLM Cost Optimization
• Service Architecture Anti-Patterns
• High-Throughput Prediction Serving
• Near-Real-Time Data Pipeline
• CLI AI Coding Assistants
• Reinforcement Learning from Human Feedback
• External Consistency
• Apache Iceberg

Lớp tổng hợp

• Scalable Distributed Systems Patterns
• API Design Patterns
• Modern Web Request Architecture
• System Design
• LLM
• AI Engineering Systems from RAG to Agents
• Production LLM System Design
• Coding Agent System Design
• Production AI Evaluation and Observability
• AI Search and Recommendation Systems
• Semantic Feed Retrieval Systems
• Multimodal and Recommendation AI Systems
• Observability for Distributed Systems
• Database Internals Tradeoffs
• Database Performance Tradeoffs
• Distributed Data Consistency Patterns
• Resilience Failure Control Patterns
• Netflix Streaming and Workflow Architecture
• Netflix Data Platform Patterns
• Netflix Java Runtime Architecture
• Kubernetes Platform Patterns
• Messaging and Event Streaming Patterns
• Object and Key-Value Storage Patterns
• Reliability Operations Loop
• Authorization and Identity Infrastructure
• Payment and Financial Data Systems
• Redis and Distributed Caching Patterns
• Deployment and CI-CD Release Strategies
• Container and Service Architecture Tradeoffs
• Search Infrastructure Patterns
• Cloud Outage Anatomy Patterns
• Production Agent Platform Patterns
• Data Platform Processing Patterns
• Serverless and Edge Runtime Patterns
• Debugging and Incident Intelligence Patterns
• LLM Threat Model and Agent Security
• Legacy Modernization and Code Migration Patterns
• Client Experience and Frontend Platform Patterns
• Multi-Tenant Architecture Patterns
• ML Platform and Prediction Serving Patterns

Cụm nguồn nên xử lý tiếp

• Auth/access-control at scale đã có lớp concept chính cho OIDC federation, token exchange, ABAC và Zanzibar/ReBAC.
• Reliability operations đã có lớp concept chính cho capacity planning, load testing, rollout, incident response, postmortem và disaster recovery.
• Object/key-value storage đã có lớp concept chính; có thể mở rộng thêm object-store security/compliance nếu cần.
• Messaging/Kafka đã có lớp concept chính; có thể mở rộng thêm Kafka cost optimization nếu cần.
• API/web edge đã có lớp concept chính cho request path, gateway policy, lifecycle, documentation, throttling và stateless API scaling.
• Payment/financial systems đã có lớp concept chính cho PaymentIntents, local payment orchestration, fraud decisioning và financial source of truth.
• Redis/distributed caching đã có lớp concept chính cho Redis internals, cache strategies, invalidation, Redis Streams, ZSet, HyperLogLog, distributed locks và feature-store caching.
• Deployment/CI-CD đã có lớp concept chính cho CI, delivery/deployment, pipeline, rollout strategies, feature flags, dark launch, shadow traffic, expand-contract, rollback alarms và bake period.
• Container/service architecture đã có lớp concept chính cho Docker, containerization vs virtualization, image/runtime, Linux namespace/cgroups, monolith/modular monolith/microservices/serverless và container design patterns.
• Search infrastructure đã có lớp concept chính cho inverted index, segment, indexer/broker split, tenant/cell isolation, destination-aware batching, query AST, ranking và zero-downtime reindexing.
• Cloud outage anatomy đã có lớp concept chính cho control/data plane, hidden dependencies, herd effect, kill switches, fail closed, cold read path, replication-not-backup, status page dependency và global metadata replication.
• Production agent platform đã có lớp concept chính cho orchestrator-worker agents, citation agents, effort scaling, state checkpointing, diff problem, sandboxed execution, zero-secret architecture, safe outputs, trust-boundary logging, protocol interoperability và agent eval stack.
• Data platform/data processing đã có lớp concept chính cho batch, micro-batch, stream processing, event time vs processing time, window, watermark, late data, lambda/kappa architecture, CDC bootstrap, pipeline validation, Spark on Kubernetes, remote shuffle service, data platform as code, warehouse/lake/mesh.
• Serverless/edge runtime đã có lớp concept chính cho FaaS, BaaS, serverless cold start, Lambda execution environment, Firecracker microVM, SnapStart, provisioned concurrency, cost model, Lambda Layer và Cloudflare Worker sharding.
• Debugging/incident intelligence đã có lớp concept chính cho debugging as code, runbook automation, automated RCA, analyzer chaining, dependency graph, context gathering, diagnostic agent và production state replay.
• LLM threat model đã có lớp concept chính cho indirect prompt injection, lethal trifecta, Agents Rule of Two, untrusted model output, LLM supply chain security và RAG knowledge-base poisoning.
• Legacy modernization/code migration đã có lớp concept chính cho traffic replay, state reconciliation, behavioral compatibility, codemod migration, dependency-driven migration, leaf-to-root migration, service-layer refactoring, intent-based test migration, idiomatic rewrite và dual-system operation.
• Client/frontend platform đã có lớp concept chính cho client state sync, streaming compression, delta update, passive session, mobile bandwidth optimization, notification budgeting/recommender/reranking, build critical path, mobile modularization, compiler-driven modularization, developer velocity, multi-tenancy/hostile multi-tenancy, sandboxed build execution, warm pool, design-to-code context, Code Connect, component mapping và design-system drift.
• Multi-tenant architecture đã có lớp concept chính cho isolation spectrum, tenant storage model, tenant context, noisy neighbor, cross-tenant data leak và resource quota.
• Database performance/internals đã có lớp concept chính cho query planner, execution plan, partitioning, SQL/NoSQL/NewSQL, MVCC/snapshot isolation và document-store joins.
• ML platform/prediction serving đã có lớp concept chính cho feature store offline/online, training-serving skew, feature discovery, data catalog, model self-tests, model shadowing, prediction logging, feedback loop, incremental training, feature drift/collocation, serving fanout, compute graph split, raw feature transport, model deployment reconciliation, model onboarding/managed serving integration, internal platform as product và annotation platform/debt.
• AI engineering còn lại: case study chuyên sâu nếu cần.
• Netflix case study còn lại: Java/runtime, chaos engineering lịch sử và multimodal AI/search đã có lớp concept chính; chỉ mở rộng tiếp nếu muốn phân tích sâu từng framework/model.
• Kubernetes/container orchestration đã có lớp concept chính; chỉ mở rộng tiếp nếu muốn đào sâu từng operator hoặc production incident.