HTTP Learning Hub
Full HTTP reference: header checker tool, glossary (HTTP/2, HTTP/3, QUIC, HPACK), how-to (CI/CD, secrets, observability), 100+ how-to guides.
Glossary (75) all →
Edge Runtime — JS on Edge CDN
Edge Runtime: V8 Isolates, Cloudflare Workers, Vercel Edge, Deno Deploy. Low latency, no Node APIs.
Rust Web Frameworks 2026: Axum vs Actix Web vs Rocket
Axum, Actix Web, Rocket and Warp compared: Tokio async runtime, latency, middleware and ergonomics of each Rust web framework.
Cloudflare R2: S3-Compatible Storage, Pricing, No Egress
Cloudflare R2 is S3-compatible object storage with zero egress fees, $0.015 per GB-month, working with aws-cli, rclone and boto3.
P95 latency — what it is and how to measure it
P95 latency — the 95th percentile of responses. Why use it instead of average, how to compute, target thresholds for API/LLM/web.
WebAssembly Component Model · Definition & Examples
WebAssembly Component Model: WIT interfaces, portable binaries, language interop. Wasmtime, JCO, ComponentizeJS.
robots.txt — What It Is and How It Works
robots.txt is a file with instructions for search bots. User-agent, Disallow, Allow, Sitemap, Crawl-delay. Common mistakes and checks.
Materialized View — What It Is
Materialized view: precomputed query result stored as a table. Postgres, Oracle. Refresh strategies.
gRPC — What It Is, vs REST
gRPC is a high-performance RPC protocol from Google. Protobuf, HTTP/2, streaming. Difference from REST.
IndexedDB — Browser NoSQL Database
IndexedDB — browser-built-in NoSQL DB. Async, transactional, large storage. For PWA and offline-first.
SSE — Server-Sent Events · Definition & Examples
SSE (Server-Sent Events) — server-to-client streaming over HTTP. Simpler than WebSocket, works through any proxy.
Vector Database — What It Is
Vector DB: optimised DB for ANN search on embeddings. Qdrant, Pinecone, Weaviate, Milvus, pgvector.
WebGPU — GPU API for the web
WebGPU: WebGL replacement, compute shaders, WGSL, access to Vulkan/Metal/D3D12. Chrome 113+, Firefox 141+.
Serverless Containers — Cloud Run, Fargate, Fly · Definition & Examples
Serverless containers: Cloud Run, ECS Fargate, Fly Machines. Container instead of function, scale-to-zero, no cold start.
CDC — Change Data Capture · Definition & Examples
Change Data Capture: stream DB changes in real-time. Debezium, Kafka Connect, WAL, binlog.
Terraform — Infrastructure as Code · Definition & Examples
Terraform: declarative IaC from HashiCorp. HCL, providers, state, modules. BSL license change 2023.
RAG (Retrieval-Augmented Generation) · Definition & Examples
RAG — LLM technique with vector DB search + context injection. LlamaIndex, LangChain, Pinecone/Qdrant.
Tool Calling (Function Calling) in LLM
Tool Calling: LLM invokes external functions via structured JSON. OpenAI, Anthropic, Gemini API.
LLM Fine-tuning — Supervised + LoRA
Fine-tuning: continued training of an LLM on your own data. LoRA, QLoRA, full FT. OpenAI, Together, Hugging Face.
LLM Quantization — INT8/INT4
Quantisation: compressing LLM to low-bit weights. GGUF, GPTQ, AWQ. Running 70B on consumer GPU.
Context Window — What It Is in LLM
Context window: max tokens the model remembers. Claude 1M, Gemini 2M, GPT-5 1M. Long context vs RAG.
AI Agent — Autonomous LLM + Tools
AI Agent: LLM + memory + tools + planning. ReAct, AutoGPT, Claude Agent SDK, LangGraph.
MoE (Mixture of Experts) — Sparse LLM · Definition & Examples
MoE: sparse transformer with expert routing. Mixtral, DeepSeek, GPT-4 internals. 1.8T params but 400B active.
Transformer — LLM Architecture
Transformer: attention-based neural network. Self-attention, multi-head, positional encoding. Basis of GPT, BERT, LLaMA.
Apache Avro — Schema-first Data Format · Definition & Examples
Apache Avro: row-oriented data format for Kafka, streaming. Schema registry, backwards compatibility.
Semantic Search — Meaning-based Search
Semantic search: search via embeddings + ANN instead of keyword match. Hybrid search. Elasticsearch, Qdrant.
React Server Components (RSC) — What It Is
React Server Components: server-only rendering, zero JS bundle to client. Next.js 13+, async components, streaming.
Event-Driven Architecture — EDA · Definition & Examples
Event-driven architecture: services communicate via events. Kafka, RabbitMQ, pub/sub. Loose coupling.
Data Lake vs Data Warehouse · Definition & Examples
Data Lake: raw data (Parquet, JSON) in object storage. Warehouse: structured + optimised for queries (Snowflake, BigQuery).
Vector Embedding — What It Is
Vector embedding: numeric representation of text/image/audio in N-dim space. OpenAI, Cohere, jina, bge.
dbt — Data Build Tool · Definition & Examples
dbt: SQL-based transformation layer in the modern data stack. Ref(), tests, docs, lineage. Jinja templating.
LLM (Large Language Model) — What It Is
LLM — large language model. GPT-5, Claude Opus 4.7, Gemini 2.5, Llama 3. Architecture, inference, sizes.
GitOps — Declarative Deploy via Git
GitOps: infrastructure + app state in Git, automated sync. Flux, ArgoCD. 4 principles.
Envoy — Proxy + Service Mesh Data Plane · Definition & Examples
Envoy: high-performance L7 proxy from Lyft. Backbone of Istio, AWS App Mesh. Configuration, filters.
Kubernetes Operator — Custom Controllers · Definition & Examples
K8s Operator: custom controller + CRD for stateful apps. Postgres, Redis, Kafka operators. OperatorHub.
Apache Parquet — Columnar Storage Format · Definition & Examples
Apache Parquet: columnar format for data lake. Compression, column pruning, predicate pushdown. Used by Spark, Trino, DuckDB.
Istio — Service Mesh for Kubernetes
Istio: most-featured service mesh. Envoy sidecars + istiod control plane. mTLS, traffic shaping, observability.
Saga Pattern — Distributed Transactions
Saga pattern: long-running distributed transactions via events, compensation. Choreography vs orchestration.
Jaeger — Distributed Tracing · Definition & Examples
Jaeger — open-source distributed tracing (Uber, CNCF graduated). Spans, service map, architecture.
Apache Iceberg — Lakehouse Table Format
Apache Iceberg: ACID table format for data lake. Netflix origin, Snowflake/BigQuery/Databricks support 2024+.
OCI Image — Open Container Initiative Format · Definition & Examples
OCI: open container image standard. Docker-compatible. Manifests, layers, registries.
Idempotency in APIs — What It Means
Idempotency — a property where an operation produces the same result on N executions. Idempotency-Key in POST, GET vs POST, patterns.
Shadow DOM — Web Components Encapsulation
Shadow DOM: scoped DOM tree + CSS for Web Components. Does not leak outer styles, used in React Native Web, Ionic.
Import Maps — JS without a Bundler
Import Maps: browser-native mapping of bare specifiers to URL. npm without webpack, dev without a bundler, CDN imports.
OpenTelemetry — Observability Standard · Definition & Examples
OpenTelemetry (OTel): vendor-neutral standard for metrics, logs, traces. CNCF. Auto-instrumentation.
HNSW Index — ANN for Vectors
HNSW: Hierarchical Navigable Small World graph. Fastest ANN index for vector search. Parameters M, ef.
Web Streams API — ReadableStream · Definition & Examples
Web Streams API: chunked data processing in browser + Node.js. Fetch streaming, compression, crypto, SSE.
WASI — WebAssembly System Interface · Definition & Examples
WASI: POSIX-like API for WebAssembly outside the browser. Server-side WASM, Fastly, Fermyon, plugins.
Edge Computing — What It Is and Why It Matters
Edge computing runs code on the nearest point (edge) to the user. Cloudflare Workers, AWS Lambda@Edge, Deno Deploy. Different from CDN.
Observability — 3 Pillars · Definition & Examples
Observability: metrics, logs, traces. Difference from monitoring. OpenTelemetry standard.
SRE — Site Reliability Engineering · Definition & Examples
SRE: Google-originated discipline. Toil, error budgets, SLO-driven. Differs from DevOps.
Log Aggregation — Centralising Logs
Log aggregation: collect logs into central store. ELK, Loki, Splunk, CloudWatch. Architecture, cost.
Service Mesh — What It Is, Istio, Linkerd
Service Mesh: infrastructure layer for service-to-service communication. Istio, Linkerd, Consul. mTLS, retries, observability.
Helm — Kubernetes Package Manager
Helm: templating + packaging Kubernetes manifests. Charts, values.yaml, Helm Hub, OCI registries.
Sidecar Pattern — Container Architecture
Sidecar pattern: auxiliary container alongside the main one. Logging, proxies, config reload. K8s multi-container pods.
Consistent Hashing — Distributed Sharding · Definition & Examples
Consistent hashing: distribute keys across nodes with minimal reshuffle on scaling. Ring, virtual nodes, rendezvous.
ArgoCD — GitOps for Kubernetes
ArgoCD: declarative GitOps continuous delivery. App-of-apps pattern, sync, rollback. UI + CLI.
Webhook — What It Is and How It Works in 2026
A webhook is an HTTP POST callback from a service to your URL on an event. Difference from polling, HMAC signing, best practices.
Canary Release — Gradual Rollout
Canary release — roll out a new version to % of traffic. Difference from blue-green, feature flags. Prometheus alerts.
WebSocket — Real-Time Bidirectional · Definition & Examples
WebSocket (RFC 6455) — a full-duplex connection over TCP. Real-time chat, notifications, games. Difference from HTTP.
LRU Cache — Least Recently Used · Definition & Examples
LRU cache: eviction policy. Redis allkeys-lru, HashMap+DoubleLinkedList. O(1) implementation.
FinOps — Cloud Cost Management · Definition & Examples
FinOps: practice of optimising cloud spend. FinOps Foundation, principles, RI/Savings Plans.
Streaming SSR — Progressive Rendering
Streaming SSR: server sends HTML in chunks as they become ready. Suspense, Transfer-Encoding: chunked, TTFB drops.
Token Bucket — Rate Limiting Algorithm
Token bucket — rate limiting algorithm with burst. Difference from leaky bucket, sliding window, fixed window.
Partial Hydration — Selective JS Activation
Partial hydration: only interactive components get JS event handlers. React Server Components, Astro, Marko.
View Transitions API — Smooth Page Transitions
View Transitions API: DOM state changes with native animations without JS libraries. Chrome 111+, same-document + cross-document.
CQRS — Command Query Responsibility Segregation · Definition & Examples
CQRS — separation of read and write models in architecture. When to use, with Event Sourcing, examples.
Event Sourcing — What It Is
Event Sourcing — architectural pattern: store events, not state. EventStore, Kafka, replay. Pros and cons.
Turbopack — Rust Bundler for Next.js
Turbopack — Webpack successor by Vercel. Rust, incremental, 10x faster. Stable in Next.js 15+.
Islands Architecture — Partial Hydration · Definition & Examples
Islands architecture: static HTML + interactive JS "islands". Astro, Fresh, Qwik. Minimal JS payload.
CRDT — Conflict-Free Replicated Data Types · Definition & Examples
CRDT — data structures for real-time collaborative editing. Yjs, Automerge, offline-first without server-mediated merge.
Feature Flags — Runtime Toggles
Feature flags (toggles): turn features on/off without deploy. LaunchDarkly, GrowthBook, Unleash. A/B testing.
SLI / SLO / SLA — Differences
SLI (Service Level Indicator), SLO (Objective), SLA (Agreement) — foundations of reliability engineering. 99.9% examples.
Bloom Filter — Probabilistic Data Structure · Definition & Examples
Bloom filter — data structure for fast membership check with false positives. Redis, CDN, Cassandra.
Blue-Green Deployment · Definition & Examples
Blue-Green deployment: 2 identical envs, instant switch. Rollback in a second. DB migrations challenge.
GraphQL — What It Is
GraphQL — a query language for APIs. Single endpoint, client requests only the fields needed. Difference from REST.
How-to (32) all →
How to Set Up an nginx Reverse Proxy — 2026
Step-by-step nginx reverse proxy setup. proxy_pass, headers, X-Real-IP, load balancing, WebSocket upgrade.
How to Set Up Prometheus + Grafana — 2026
Install Prometheus + Grafana: Docker compose, scrape config, data source, dashboards, Node exporter.
How to Set Up a CDN — Cloudflare, AWS, Yandex 2026
Step-by-step CDN setup: Cloudflare free, AWS CloudFront, Yandex Cloud CDN. DNS, SSL, cache rules.
How to Set Up PostgreSQL Backups — 2026
Step-by-step PostgreSQL backup: pg_dump, pg_basebackup, WAL archiving, point-in-time recovery (PITR). S3 storage.
How to Create a systemd Service — 2026
Step-by-step systemd unit file for Node.js/Python/Go apps. Auto-start, restart policy, logs, security.
How to Deploy a Vector Database — 2026
Qdrant, pgvector, Weaviate setup. Docker deploy, collection config, HNSW tuning, monitoring.
How to Configure logrotate — 2026
Step-by-step logrotate setup for application logs. Daily rotation, compression, retention, post-rotate hooks.
How to Stream LLM Responses in Real Time
Server-Sent Events, streaming LLM answers in browser. OpenAI, Claude, Gemini. Fetch API + ReadableStream.
How to Add Semantic Search to a Site — 2026
Embeddings-based search for blogs, docs, e-commerce. OpenAI + Qdrant + native UI.
How to Set Cache-Control Headers — 2026 Guide
Proper Cache-Control, ETag, Last-Modified setup for static assets and HTML. nginx and Apache examples, TTL per content-type.
How to Generate SBOM for Supply Chain
SBOM (Software Bill of Materials): syft, CycloneDX, SPDX. CISA compliance, executive order 14028, sign + attest with cosign.
How to Set Up OpenTelemetry — 2026
OpenTelemetry (OTel): standard for traces + metrics + logs. Setup Node/Python + collector + backend (Jaeger/Grafana/Datadog).
How to Trace Distributed HTTP/gRPC Calls
Distributed tracing via OpenTelemetry: traceparent headers, context propagation, span correlation. Jaeger, Tempo, Honeycomb.
How to Measure SLI/SLO for a Service
SLI + SLO + error budget: SRE practices. Latency p99, error rate, availability. Alert when budget exhausted.
How to Set Up Prometheus Alerting — Alertmanager
Prometheus Alertmanager: rules, routing, deduplication, PagerDuty/Slack/Email integration. Best alerting practices.
How to Set Up Structured Logging — JSON Logs
Structured logging: JSON format, correlation IDs, log levels, Pino/Winston/Zap. Elasticsearch/Loki/Datadog backends.
How to Find and Fix Slow Postgres Queries
PostgreSQL performance: pg_stat_statements, EXPLAIN ANALYZE, indexes, N+1 queries, connection pooling.
How to Use Docker Compose — 2026
Docker Compose: multi-container apps via docker-compose.yml. Networks, volumes, env files, scaling, healthchecks.
How to Debug nginx Errors — 2026
nginx error debugging: error.log, -t config test, debug module, 502/504 causes, permissions.
How to Set Up Cloudflare CDN for a Site — 2026
Cloudflare setup: DNS, Proxy (orange cloud), SSL/TLS, Page Rules, cache, security. Free tier setup.
How to Fine-tune LLM on Your Data — 2026
Fine-tune LLM via LoRA / OpenAI / Together.ai. Prepare JSONL dataset, train, eval, deploy.
How to Set Up CI/CD in GitHub Actions — 2026
Step-by-step CI/CD in GitHub Actions: workflow YAML, jobs, matrix, secrets, deploy to server. Full example.
How to monitor LLM API latency (OpenAI, Anthropic, Yandex)
LLM API latency monitoring: P50/P95/P99, streaming vs full responses, time-to-first-token, retries and Slack/Telegram alerts.
How to monitor API uptime (HTTP check, status codes, timeouts)
API uptime: proper HTTP checks, expected codes, timeouts, retries, multi-region, false-alert filtering, Slack/Telegram integration.
How to Set Up Redis Cluster — 2026
Redis Cluster setup: 6 nodes, sharding, replication. Automated failover, client-side routing. Production guide.
How to Build a RAG Chatbot on Docs — 2026
Step-by-step: embeddings → vector DB → retrieval → LLM. LangChain, Qdrant, OpenAI. Full example in 30 min.
How to Cache LLM API Calls — 2026
Redis cache for LLM responses. Exact match + semantic cache. Anthropic prompt cache. 10x cost reduction.
How to Evaluate LLM Output Quality — 2026
Ragas, LangSmith, LLM-as-judge. Automatic metrics for RAG, chatbot, code generation. CI/CD eval pipelines.
How to Build an AI Agent with MCP — 2026
Model Context Protocol (MCP) from Anthropic: expose tools to LLM. Claude Desktop, Zed IDE, custom servers.
How to Deploy LLM on Serverless GPU — 2026
Modal, RunPod, Replicate, Cloudflare Workers AI. Serverless GPU for LLama 3, Mistral, custom models.
How to Set Up GitOps with ArgoCD — 2026
Step-by-step GitOps with ArgoCD: install, first Application, auto-sync, multi-cluster, app of apps.
How to Migrate an Application to Kubernetes — 2026
Step-by-step migration from VPS/Docker Compose to Kubernetes: containerization, manifests, Helm, observability.
Alternatives (6) all →
Host-Tracker Alternative: Bigger Free Plan, API Included
Host-Tracker and a bigger free plan: check a URL from 3 regions here, then monitor it. Host-Tracker wins on alerts: SMS, voice, Viber.
updown.io Alternative: Flat Plans, No Credits, Free Tier
updown.io credits vs flat monthly plans: check a URL from 3 regions here, then monitor it free. updown.io charges one credit per check.
Downdetector Alternative: Monitor Your Own Site Free
Downdetector and a monitor for your own site: Downdetector counts user reports for big services; here you check your own URL from 3 regions.
Datadog Alternatives 2026 — Observability
Datadog — enterprise APM + infrastructure monitoring. Alternatives: Enterno.io, New Relic, Grafana Cloud, self-hosted Prometheus.
New Relic Alternatives 2026 — APM
New Relic — APM + infrastructure monitoring. Alternatives: Enterno.io, Datadog, Grafana, Elastic APM.
Oh Dear Alternatives — Application Monitoring 2026
Oh Dear is full-stack site monitoring. Alternatives: Enterno.io, BetterUptime, Checkly, UptimeRobot.
SSL errors (4) all →
ERR_EMPTY_RESPONSE: How to Fix It (Chrome, Edge)
The server closed the connection without replying: nginx return 444, a crashed app or HTTP sent to port 443. Reproduce with curl -v.
ERR_HTTP_RESPONSE_CODE_FAILURE (-379): How to Fix It
A 4xx/5xx reached a Chromium caller that reads the status: empty error page, WebView, PAC or cert fetch. Find the URL in net-export.
ERR_INCOMPLETE_CHUNKED_ENCODING: How to Fix It
The body was cut before its last chunk, usually by proxy_read_timeout or a backend killed mid-response. Check the nginx error log.
NET::ERR_INVALID_HTTP_RESPONSE — Malformed Response · Browser Error Fix
Chrome ERR_INVALID_HTTP_RESPONSE — server returned HTTP data that does not conform to HTTP/1.1 RFC. Proxy corruption, content-length mismat…
Research (17) all →
Nginx Market Share 2026: Why Surveys Say 14% to 44%
nginx holds 14.45% to 43.92% in Netcraft's July 2026 survey and 31.3% at W3Techs. Why the numbers differ, and 85.9% of .ru.
LLM shares 2026: four metrics, four winners
By human visits OpenAI leads (~53%), by crawler traffic Google (24.8%), among developers Anthropic (39%). None of the answers is wrong. And…
Headless CMS 2026: the share is structurally unmeasurable
Our measurement of 401 pages found headless systems on 11 — 2.7% against a third for traditional ones. But that is not a share; it is the s…
AI in the browser 2026: 84% can, 0.004% do
WebAssembly is supported by 95.96% of browsers and WebGPU by 83.99%. Yet GPU work is submitted on 0.003967% of page loads. A gap of roughly…
RAG in 2026: pgvector alone equals every vector database
pgvector draws 37.3M downloads a month against 38.0M for Qdrant, Chroma and Pinecone combined. In JavaScript the Vercel toolkit outdraws La…
Russian AI services 2026: a report on 30,449 production calls
YandexGPT’s median response is 3,793 ms, with 0.14% empty responses across 30,449 calls. Generation speed holds at about 14 ms per token an…
AI cost 2026: caching removes half the requests
Across three months 48.6% of model calls were served from cache and never reached a provider — a share that rose from 23% to 52%. With an i…
DB drivers 2026: PostgreSQL leads both npm and PyPI
PostgreSQL leads in JavaScript (64.6M downloads a week) and Python (530.4M a month). MongoDB outranks MySQL in JS and trails it in Python. …
npm dependencies 2026: half of packages have none at all
Four independent samples: 49–56% of npm packages have no dependencies at all, with a median of zero. TypeScript resolves 21 packages and in…
Response times 2026: median 95 ms, p95 at 966 ms
Across 1,395,771 successful checks: median 95 ms, p95 966 ms — a tenfold spread. Ping answers in 16 ms, HTTP in 96, API in 153. The p95 spr…
Image sizes 2026: python at 393.7 MiB against alpine’s 16.1
Our Docker Hub API measurement of 27 Aug 2026: python:3.13 weighs 393.7 MiB, its alpine variant 16.1 — a 24-fold gap. Growth across major v…
Container runtimes 2026: no measurement of shares exists
The last runtime telemetry is Datadog’s November 2023 report (containerd 53%, Docker 65%) and the series is discontinued. CNCF’s September …
AI in development 2026: 84% use it, 3.1% trust it
Stack Overflow 2026 survey results do not exist — figures labelled "SO 2026" are 2025 data. The one genuine 2026 measurement (JetBrains, Ma…
CSS frameworks 2026: two thirds of sites use none
A 399-page measurement: Bootstrap found on 20–28% of sites, Tailwind on 3% in .ru against 13% in .com. Two thirds of sites in every zone sh…
Bot share 2026: 35.5% of requests or 61.9% of pages
Across all HTTP requests bots account for 35.5%; across HTML page requests, 61.9%. Our 600-host measurement: about one site in ten disallow…
JS frameworks 2026: React on 6.1% of sites or used by 85% of devs
React was detected on 1,306,221 origins (HTTP Archive, July 2026), on 6.1% of sites (W3Techs), and named by 85% of State of JS respondents.…
WordPress and CMS in 2026: 40.7% of sites or 58.9% of the market
WordPress runs on 40.7% of all sites and holds 58.9% of the CMS market — different quantities. In the .ru zone: WordPress 33.5%, Tilda 27.1…