PSA: Stop starting startups on Firestore
engineering.traba.workNot sure if this company also ran into this issue, but the engineers at my startup absolutely hated the Firebase UI for querying. Genuinely could not fathom how product designers could put together such a poor experience. Needless to say we did our own migration after 1 year to MySQL after
This reminds me of that April Fools tweet on Google sunsetting Firebase that I fell for lol
lol this would actually happen tho
Pretty incredible how Postgres has gotten so dominant in eng mindshare over the last 5-8 years. Especially with super popular tools like Supabase now being built just on top of Postgres; feel like every company I see is on it
good article, enjoyed the little storytelling even if a bit distracting. i think most folks on here are going to be super critical about them starting on a NoSQL database in the first place when it was very likely that relational would have been the way to go from the start, but sometimes you just need to get something off the ground and Firestore can be the fastest to do so (not recommending fwiw). we were also one of those “thousands” of startups that started with it and migrated off after a few years too - so might be a bit biased there - but it all worked out in the end
lol I agree re:storytelling - thought it was a nice touch. gave the article a bit of flair that was more interesting to read than many eng blog posts that can be otherwise pretty dry
Document dbs, like crypto cycles, keep coming in and out of the zeitgeist...why? I just don't understand how purportedly smart engineers keep falling for the same traps over and over
That’s offensive to the innovation happening in crypto to be honest, compared to the lack of it in the non-relational database world
What is even left to innovate in document databases? They do a very specific thing well; it's not their fault naive engineers keep trying to use them in ways they are not intended to be used
mistakes will always be made at all stages, imo the key differentiator is how quickly they course correct and iterate
It is not a mistake when it is a decision made after a well-thought tradeoff evaluation.
Tech startup with very relational data picks hyped up non-relational database and regrets decision later… classic
if everyone committed to using tried-and-tested things that have worked for half a century we wouldn’t need to waste so many man hours on these types of tasks