For a technology with such obvious upside, serverless databases are still oddly… niche. Market research firms estimate the serverless database market at ~15-20% of the broader cloud database market. In other words: plenty of teams are still defaulting to provisioned capacity even when their demand is anything but predictable. Meanwhile, “serverless” more broadly has gone mainstream everywhere… except where the data lives.
So what are we leaving on the table? The core promise is simple: stop guessing capacity and optimize costs! Modern serverless relational offerings can auto-scale with workload demand, charge per-second based on consumed capacity, and—even more provocatively—scale down to zero when idle, reducing overprovisioning and “always-on” waste.
In this talk, I’ll walk through the real reasons teams hesitate (predictability, performance, operability, and “database physics”), then share the practical engineering work we’re doing to make Azure HorizonDB Serverless real.