Worst Case Optimal Joins: Graph-Join Correspondence

Databases just went full detective board: this post maps old-school SQL joins (how tables connect) to graph patterns (think dots and lines), and even shows how to find triangles — three nodes all connected — with a cleaner Datalog style. The vibe? Graph folks are cheering, SQL lifers are rolling their eyes, and performance hawks are screaming, “Show me the benchmarks!” One commenter, eatonphil, dropped receipts that this is part of a series, which only fueled the binge-watch energy.

Fans love that “worst-case optimal joins” (a mouthful meaning a method designed to stay fast even in tough cases) makes joins feel like patterns you can see, not math you fear. The Datalog example using EAV triples (storing facts as tiny three-part entries) sparked a split: some called it cleaner, others roasted it as “hipster SQL.” Meanwhile, the triangle-hunting demo inspired memes like “Triangles in my database? More likely than you think.”

Drama peaked around whether you need a special graph database at all; a loud crew insists you can model graphs in regular SQL just fine, while graph DB fans say, “Try that at scale and call us back.” Another faction worried the self-join party means “your cloud bill will join too.” And when the author teased graph problems like vertex cover, half the thread had math class flashbacks — the other half asked for charts and numbers, now.