Today it's exactly ten years since I published my blog post called Mythbusters: Stored Procedures Edition.
Using "stored procedures" (ie storing and executing code in your database, next to the data, instead of in a middle tier or client) is a practice that is provably both efficient and secure, yet it is often dismissed as "wrong".
My original blog post looks at the usual reasons given against the use of stored procedures for "business logic" (or for anything at all, really) and explains why they tend to be myths (or outright lies), repeated so many times that they are taken as the truth. Take a moment to read the article now.
Publishing the article proved that advocating for the use of stored procedures, which had gone out of fashion already by 2011, was controversial. It caused an immediate spike in my blog traffic, receiving almost 20 000 page views in a single day (at the time, my blog had perhaps a few hundred page views per day on average). The article was picked up by Hacker News and generated a lively discussion there.
So, where are we in 2021, ten years later?
Mainstream development (still) uses imperative, object-oriented 3GL languages (Java, .NET, etc) running in the middle tier or client, dragging data out from the database to process it, with data access usually abstracted through an ORM. Which, predictably, results in suboptimal performance.
The myths about stored procedures are still prevalent. Ask a random Java/.NET developer today, and I bet he will claim that it's impossible to version control stored procedures (myth #1). In that case, please send him a link to that 10-year old article...
