Checking to see if anyone is using Replit for customer facing APIs. I tried looking at documentation and I could find compute costs, but I’m looking for rate limiting/load data. I know most companies have policies that prevent you from load testing, but I’m just trying to get some basic data around volume.
Scenario:
Create records (couple thousand, bulk would be ideal but single record would be fine)
Update records (less but could still be in thousands in some scenarios)
I’d also be ok with only flat-file uploads to start and still need to support a few thousand record creates, and would need to support bulk and the other form of multiple calls in a single request that is banned from being mentioned on this forum, in a year or so.
Anyone know the limits or has anyone done customer facing yet?
I’m building an ingestion and distribution layer for a large diamond inventory (~600k records) for https://izios.com intended to sync into Shopify and WooCommerce through plugins as well as API. At that scale, direct per-record API reads are not practical i’ll either overwhelm my database or blow through read/write limits (especially on managed Postgres like Neon).The cleanest approach is bulk ingestion (CSV or equivalent snapshots), combined with hashing to detect deltas so only changed records are processed; that lets me avoid unnecessary writes and keeps the system stable.
On the serving side, the data is cached and exposed as normalized JSON for downstream consumers (storefronts, search, filters), so the API layer isn’t hitting the database on every request. This keeps API usage predictable, prevents database thrashing, and scales cleanly as inventory grows.
Thanks for sharing! 2k records would be just the initial ingestion, so significantly less volume, but I didn’t think to try and check Neon/Postgres limits. My main want for external APIs is more for upsert and allowing customers to keep records up-to-date (mainly Users because there’s high turnover in Property Management roles and their management). I have CSV already supported, but I want the platform to feel more intuitive than just supporting flat-file.
Your use case sounds amazing though, that’s some high volume! I saw some use case for Airtable where they auto-generated all metafields/metaobjects, looked like a cool way to make sure all info was associated with high volumes like this. Very interesting.