PWV Portfolio Feed

Thanks Apple for doing more open source 😀. These contributions wouldn’t be possible otherwise.

there was a secondary market for startup equity that used graphql and just dumped their whole data model in there, the names and emails of all sellers and buyers were just there in plaintext on every request to the page

✏️ Conclusion

The hard truth about query performance is that it's possible to write fast as well as slow queries with each ORM library!

In the end, much of an application's performance depends on the developer's ability to follow best practices, identify slow queries, and optimize them over time.

🔎 If you zoom out and look at the results from a distance, you'll notice that most queries actually perform in similar ballparks with only a few milliseconds difference. 

As an example, here are the results we've collected on AWS RDS:

🤔 So, which ORM is the fastest?

As unsatisfying as it may be, the answer is as so often: It depends! Performance is a nuanced topic and depends on lots of factors.

While we can't provide a conclusive answer, analyzed some patterns in the blog post (link at the end).

⏳ Collecting data

We collected data by executing the benchmark scripts 500 times against various databases.

We intentionally made the repo open-source — feel free to give it a spin yourself and share your results with us!

🏗️ Infrastructure

We executed the queries from an EC2 instance and ran them against different Postgres instances of various providers.

📊 Last summer, we wanted to know how Prisma ORM compares to other ORMs in performance and created open-source benchmarks!

So, which ORM is the fastest? The (maybe unsatisfying) answer: It depends!

Read on to learn about our methodology and results ...

confession: I hedge my bets by making sure I stay friendly with Claude, just in case he wakes up.

Call it AI-simping. I call it the Pascal AI-wager.