Every article tagged Performance, in one place.
Showing 29 articles
Any list endpoint needs pagination. The two main approaches behave very differently on large or fast-changing data. Offset pagination GET /articles?page=3&per_page=20 SELECT * FROM articles ORDER BY published_at DESC LIMIT 20 OFFSET 40; Pros: simple, lets user...
Installing dependencies is often the slowest part of a CI run. Caching downloads between runs can cut minutes from every workflow. The easy way: built into setup actions - uses: actions/setup-node@v4 with: node-version: 22 cache: npm # also: yarn, pnpm - uses:...
Indexes are the single biggest lever for query performance. The right one turns a full table scan into a lookup of a few pages; the wrong ones slow down every write for nothing. B-tree: the default CREATE INDEX orders_customer_idx ON orders (customer_id); Work...
Most slow MySQL queries are missing an index or can't use the one that exists. EXPLAIN shows you which. Find slow queries SET GLOBAL slow_query_log = 'ON'; SET GLOBAL long_query_time = 0.5; -- seconds Or query performance_schema.events_statements_summary_by_di...
A cache in front of a slow database query or API can cut response times from hundreds of milliseconds to one. Getting freshness and failure behaviour right is the hard part. Cache-aside (lazy loading) async function getProduct(id) { const key = `cache:product:...
Caching in Next.js used to be implicit and confusing. Recent versions moved to an explicit model: data is fetched fresh by default, and you opt in to caching with the 'use cache' directive. Turn on Cache Components // next.config.ts import type { NextConfig }...
EXPLAIN shows how PostgreSQL plans to run a query. EXPLAIN ANALYZE actually runs it and adds real timings. It's the most important tool for fixing slow SQL. Run it EXPLAIN (ANALYZE, BUFFERS) SELECT o.id, o.total_paise FROM orders o WHERE o.customer_id = 42 AND...
Without an index, MongoDB scans every document in a collection ( COLLSCAN ). Indexes make queries fast, and compound index order matters as much as in SQL. Creating indexes await orders.createIndex({ customerId: 1 }); await orders.createIndex({ status: 1, crea...
Smaller images pull faster, start faster, cost less to store and contain fewer packages with vulnerabilities. Going from 1.2 GB to under 200 MB is common. 1. Choose a smaller base image node:22 (Debian full) - large, has build tools. node:22-slim - Debian with...
Many apps send the same large prefix on every request: a long system prompt, tool definitions, or a document the user is asking about. Prompt caching stores that prefix so later requests read it from cache at a fraction of the normal input price and with lower...
When a request needs several independent lookups ("compare the weather in Delhi, Mumbai and Bengaluru"), the model can return multiple tool_use blocks in a single response. Running them one after another wastes time. Execute in parallel const toolUses = respon...
Reading a 5 GB log file with fs.readFile loads all 5 GB into memory. Streams process data in small chunks, so memory use stays flat no matter how big the input is. The four stream types Readable - a source (file, HTTP request, database cursor). Writable - a de...
Requests and limits tell Kubernetes how much CPU and memory a container needs and may use. They affect scheduling, performance and stability, and getting them wrong is a top cause of mysterious slowdowns and crashes. resources: requests: cpu: 250m # 0.25 of a...
Short answer: useMemo caches a computed value , useCallback caches a function , and React.memo skips re-rendering a component when its props haven't changed. const visible = useMemo(() => products.filter((p) => p.inStock), [products]); // value const onSelect...
Re-renders in React are cheap until they aren't. Developers used useMemo , useCallback and React.memo to skip unnecessary work, but getting the dependency arrays right is tedious. React Compiler is a build-time tool that adds this memoization automatically. Wh...
Node.js runs your JavaScript on a single thread. That's fine for I/O, but a CPU-heavy task (resizing images, parsing a huge file, heavy maths) blocks the event loop, and every other request waits. Spot the problem app.get('/report', (req, res) => { const resul...
Traffic changes through the day. The Horizontal Pod Autoscaler (HPA) adds Pods when load rises and removes them when it falls, so you pay for what you use and stay fast at peak. Prerequisites metrics-server installed in the cluster (most managed clusters inclu...
Each PostgreSQL connection is a separate server process using memory. Hundreds of app instances each opening twenty connections will exhaust max_connections and slow the database down. Pooling fixes this. Layer 1: the application pool import pg from 'pg'; expo...
LIMIT 20 OFFSET 100000 looks innocent, but MySQL must read and throw away 100,000 rows to return 20. Deep pages get slower and slower. The problem SELECT id, title, created_at FROM posts ORDER BY created_at DESC, id DESC LIMIT 20 OFFSET 100000; -- reads 100,02...
Fetch 50 books with their authors, and a naive server runs 1 query for the books plus 50 queries for authors. That's the N+1 problem , and it's the most common GraphQL performance issue. Why it happens Book: { // called once per book, each with its own query a...
Exact nearest-neighbour search checks every vector, which gets slow past a few hundred thousand rows. Approximate nearest neighbour (ANN) indexes trade a little recall for a huge speed-up. pgvector offers two: HNSW and IVFFlat . HNSW (Hierarchical Navigable Sm...
Some events fire very often: typing, scrolling, resizing. Running expensive work (an API call, a layout calculation) on every event wastes resources and can make the page janky. Debounce: wait until things calm down Run the function only after the events have...
A list holds every item in memory. A generator produces items one at a time, on demand. For large files, streams or infinite sequences, that's the difference between working and running out of memory. yield def read_lines(path): with open(path) as f: for line...
A CI pipeline that takes 25 minutes slows down every pull request. Most pipelines can be made several times faster with a few changes. 1. Cancel outdated runs concurrency: group: ci-${{ github.workflow }}-${{ github.ref }} cancel-in-progress: true When you pus...
Images and fonts are the most common causes of slow Largest Contentful Paint (LCP) and layout shift (CLS). Next.js includes components that fix both with very little effort. next/image import Image from 'next/image'; import hero from './hero.jpg'; // static im...
A memory leak shows up as memory that keeps climbing until the process is killed or restarts. The fix starts with proving where the memory goes. 1. Confirm the leak setInterval(() => { const { heapUsed, rss } = process.memoryUsage(); console.log(`heap ${(heapU...
For decades, CPython's Global Interpreter Lock (GIL) allowed only one thread to run Python bytecode at a time. Threads helped with I/O, but not with CPU-heavy work. Free-threaded builds of Python remove the GIL. The timeline Python 3.13 introduced an experimen...
Redis lives in RAM, so memory is the resource you'll manage most. Without limits, a growing cache can exhaust the server and get Redis killed. Set a limit # redis.conf maxmemory 2gb maxmemory-policy allkeys-lru Leave headroom on the machine for the OS, client...
skip() and limit() are the obvious way to paginate, but skip(100000) still walks past 100,000 documents. Range-based pagination keeps every page equally fast. Range (keyset) pagination // First page const page = await posts .find({ status: 'published' }) .sort...
