Database Performance In The Cloud

This page contains an index to a blog series exploring the design, performance and operational realities of running enterprise databases in the public cloud.

When hyperscalers became the default destination for new workloads, the industry narrative was clear: cloud-first, cloud-only, cloud-everything. Analyst forecasts projected that the majority of databases would migrate to cloud platforms within a few short years. Infrastructure would become elastic. Capacity would become programmable. On-premises would fade into legacy status.

But mission-critical databases do not behave like generic workloads.

This series examined what actually happens when systems of record move from carefully engineered on-premises environments into shared, standardised cloud infrastructure. It explored the tension between elasticity and predictability, between abstraction and control, and between theoretical scalability and real-world performance.

Cloud did not remove the laws of physics. It simply changed where they show up.

Looking back, this series captured the second major architectural shift in the flashdba journey: not a change in storage technology, but a change in operating model. And as with storage before it, the implications for database performance were deeper — and more nuanced — than the headlines suggested.