An alternative is our self-serve on-premise product: https://retool.com/self-hosted/, which you can host in your own EC2. Given this latency issue is our fault, if you want a discount, just reach out to me; I'm david AT. Happy to provide a discount until we go multi-region with our backend.
I wanted to follow up on the above. Solving the round trip problem David mentioned above involves multi-region for our entire backend, inclusive of our DB. This is definitely a priority and something we plan on having by EoY.
Also IIRC, there was way more than 2 round trips .. I think my UI made a few (GraphQL) API calls and they appeared to be serialised, so it really was in the order of 1-2 seconds for a page to load. This was 18 months ago so both my memory and your implementation might have changed since then.
In terms of using retool, unfortunately that particular ship has sailed for me, but I’ll be quite happy to reconsider it in any of my new projects once you have a solution that is a couple of hundred light-milliseconds closer :)
Another topic of interest is managing queries. What happens if a database column is renamed for example? Now we manage it though stored queries in our db - this creates some dependencies and there is always a risk that old queries are no longer used or some query has not been updated. A single place to manage and overview all queries would go a long way (maybe this exists, I have only had a small part in the implementation).
Hopefully this is something money can fix :)
the problem was the sum of UI latency to US(?) plus latency of serial API calls to get the data. We could probably have lived with a full rendering time of 100ms or so, but responsiveness requirements don’t scale linearly. So with a few round trips, it quickly became insurmountable.