Go Full Stack Server-Side Rendering vs SPAs
summary
Compares a React SPA (100-product API) to a Go full‑stack approach using Templ + HTMX with Tailwind; JSON payload is 54 KB vs HTML ~423 KB in one test but full‑page HTML ends up only 1.8% slower (331 ms vs 337 ms total latency). Hosting on CloudFront can cut perceived load to ~30 ms if pages are cacheable, but a user‑specific cart cookie prevents caching; solution implemented is component islands that lazy‑load the cart via HTMX, enabling CDN caching and a ~10x latency win for cached pages.
AI-written summary · about
chapters (10)
- 0:00Full stack Go apps
- 3:12E-commerce demo
- 7:04SSR with Go templates
- 10:40SPA vs SSR payload
- 13:07SSR vs SPA latency
- 17:02CDN limits for SSR
- 20:36Component islands
- 24:11Rendering spectrum
- 25:58SEO for SSR vs SPA
- 27:19Questions and answers
related talks
45mSPA, isomorphic and back to the server: our journey with JS
16mThe time to go full stack is now
28mSingle Page Applications - was it a good idea?
28mServer-Side Apps, Not Server-Side Rendering: A Real Alternative to React
32mWhy you don’t need a Single Page App
45mGive Me Back My Frontend!
29mSPAs are dead. Long live Microfrontends.
31mRethinking composable E-Commerce Frontends