Tag
Every post tagged performance.
Backend & APIs
I loaded the same 200,000 order documents into MongoDB 9.0 and PostgreSQL 18 JSONB and measured inserts, indexed lookups (GIN vs B-tree), containment filters, nested updates, aggregations and disk size.
Backend & APIs
I benchmarked a hand-written SKIP LOCKED queue, pg-boss 12 and BullMQ 6 on both Redis and PostgreSQL: enqueue speed, throughput, pickup latency, crash recovery, transactional enqueue and storage per job.
Backend & APIs
I scored built-in tsvector, pg_search, pg_textsearch, Elasticsearch 9.5 and Meilisearch 1.55 on three BEIR datasets: ranking quality (nDCG@10), latency, index size, typos and how fast new rows become searchable.
AI Engineering
Tested setup for adding a remote MCP server to Cursor, Claude Code, VS Code and Claude Desktop, using a real public endpoint. Copy-paste configs, curl and Inspector checks, mcp-remote, and what tools/list costs in latency and tokens.
Next.js & React
TanStack Start 1.0 is stable, so I built the same store in Start and Next.js 16.4 and benchmarked both. Start served 2.9× to 8.6× more dynamic requests, shipped 20% less JS and built 3.9× faster. Next.js won on static pages.
Developer Tools
I ran ESLint 9 and 10, Oxlint 1.88 and Biome 2.5 on this site's real Next.js code. Oxlint was 81× faster than ESLint with the same rules and matched all 80 typescript-eslint findings 16× faster. Here's what each migration keeps and breaks.
Backend & APIs
I benchmarked Hono, Express 5, Fastify 5 and NestJS 12 across JSON, route parameters and POST bodies on Node 24, Node 26 and Bun. Hono and Fastify tie on Node at about 1.4× Express; on Bun, Hono hits 118k requests per second.
Backend & APIs
I ran the same Hono, Express and Postgres apps on Bun 1.4.3, Node.js 24 and Node.js 26, then measured throughput, memory, startup, TypeScript, installs, next build and native addons. Here is when Bun is worth switching to, and when Node 26 is enough.
AI Engineering
I built the same service as a REST API and as a stateless MCP server, then measured latency, throughput, bytes and the tokens a model reads. Here is when you need MCP, when a plain API wins, and how to wrap one with the other.
