3 ms·
Show HN: Migrated from Supabase to Pocketbase => 4x+ text search speed in my app
Hi everyone,
TLDR; built an app with large dataset and moving from supabase to Pocketbase improved speed dramatically. I recommend anyone checking Pocketbase for your next project.
I wanted to post a follow up to share some surprising results I got while building https://amznbargains.com https://amznbargains.com in public.
See full context below but in a nutshell the app collects hundreds of thousands of Amazon Warehouse deals and organize them for a better UX.
To my surprise the volume of records quickly became an increasing struggle for Supabase when performing a full-text search on the whole dataset. Even after indexing, the queries ran above 8s when not simply timing out.
Some quick prototyping on Pocketbase gave me optimistic results and after full migration I'm now averaging sub-2,000ms responses on much bigger datasets.
To be fair Pocketbase does not support real fts yet so it's not an apple to apple comparison but with my use case the simpler search it offers is totally fine to start with. Outside of text searches the overall responsiveness of the application has visibly improved with sections and filtering feeling far snappier than before.
When factoring in the absolute ease to self-host Pocketbase I'm clearly switching to it as my go-to base for starting projects.
---------
Context on the application
A web app to see, filter and sort through Amazon Warehouse deals easily. Limited to US market for now but will add coverage as I go.
https://amznbargains.com https://amznbargains.com
Why: Convenience and Affordability. This came from a personal frustration of not being able to browse efficiently across deals while on Amazon's website. They have a bazillion refurbished or like new items available but no good way to search and sort them out without clicking through individual pages. After building a prototype for my friends and family I realized most seemed to be able to find items of interest for much cheaper than they were expecting. That's when I got the idea to open up this project externally to get ideas/feedback to iterate on and hopefully help folks find cheaper alternatives before buying anything.
Tech stack:
Backend:
- Core: Nodejs via PM2 + Bash
- Workers: Proxmox VMs behind rotating VPNs
- Database: Pocketbase (was Supabase)
Frontend:
- Core: Sveltekit
- CSS: Tailwind CSS
- UI Framework: Shadcn-svelte
- Testing: Playwright
- CDN, DNS & Hosting: Cloudflare
- Analytics: Plausible
Notes on Privacy:
- No registration required to get service.
- No personal data ever stored or even less exchanged/resold.
- No ads, commercial collaboration or algo biased by monetary incentives.
- Plausible selected because it allows me to track the few metrics required to run the app while respecting stringent privacy guidelines.
- The only source of revenue will be through the Associate Program as it comes at no cost to users and requires no data to be collected/exchanged.