Sean Hammond

Unit testing your code’s performance, part 1: Big-O scaling Your software’s speed needs some form of testing; big-O scaling is a good starting point.

How to test the scalability of your code using the bigO library from UMass and pytest: pythonspeed.com/articles/big...

Oh, and no Kubernetes!

But unbundle your services--database, logs, error-tracking, auto-scaling, etc--from Heroku (or whichever PaaS you're on). It's usually cheaper and it means you're less locked-in to your PaaS. Also dedicated companies tend to provide better services than a PaaS's built-in alternatives.

TLDR: He'd choose Heroku to start with - the ease and simplicity is worth the extra costs, you've got better things to spend your time on. Heroku becomes more questionable later as usage grows, particulalry if you have high-traffic and thin margins. In that situation he'd probably switch to Fly.

Black Box Hosting vs. Glass Box Hosting: An Interview With Judoscale's Adam Founder interview comparing Heroku vs Fly/Render/Railway for bootstrapped SaaS: cost, control, portability, third-party services, simple rules.

This is a good interview with Judoscale's founder about hosting choices for new startups.

judoscale.com/blog/black-b...

Clear the functests search index when clearing the DB by seanh · Pull Request #9638 · hypothesis/h For speed h does not normally clear the DB between functests unless the test has @pytest.mark.parametrize("with_clean_db"), see: #6845 At the time when this change was made the search ind...

History lesson turned out to be incomplete: github.com/hypothesis/h...