Chapter 2:
Why now? Market forces reshaping database strategy and the rise of Kubernetes
Cloud cost scrutiny is forcing a reckoning with database strategy. According to the 2025 Flexera State of the Cloud Report, organizations estimate that 27% of their cloud spending is wasted on resources delivering no business value, and that waste isn't distributed evenly.
This problem is compounded by the fact that this spending is also growing at an explosive rate, forcing a shift in executive thinking. Forrester notes that organizations now require a “total-cost-of-ownership view that spans... platform-as-a-service solutions,” and that traditional cost tools are “getting quickly outdated” because they cannot handle this complexity.
With their opaque pricing models (which bundle hidden management fees with unpredictable I/O and data transfer costs) and vendor lock-in, database services represent some of the largest line items in cloud bills and some of the hardest costs to control.
The same scrutiny is hitting AI workloads. Analysis by Deloitte highlights a “public-cloud cost cliff” for AI: beyond an inflection point (around 60–70% of dedicated TCO), private or hybrid options become more cost-effective. That pressure spills into the data layer. Consolidating databases on Kubernetes keeps data gravity close to AI pipelines, reduces latency and egress, and preserves portability across clouds and on-prem.
CIOs facing pressure to cut cloud waste are discovering that proprietary DBaaS platforms, while operationally convenient, are financially expensive in ways that only become clear at scale. Data transfer fees, multi-region replication costs, and vendor-specific features create lock-in, making optimization impossible. This realization is driving organizations toward an alternative platform: running their databases on Kubernetes. And for many of them, they are finding that open source solutions on this platform provide the same operational benefits without vendor markup.
Kubernetes reaches critical mass
Kubernetes adoption for databases has reached an inflection point. The 2024 DoK Report shows that nearly half of organizations now run 50% or more of their data workloads on Kubernetes in production. For the third consecutive year, databases rank as the number one workload type on the platform. What was experimental three years ago is now standard practice.
The economics explain the shift. Proprietary DBaaS platforms carry an implicit "management tax" of 80-100% or more over the cost of the underlying IaaS resources, as the TCO analysis in Chapter 5 demonstrates. As cloud budgets tighten, CIOs must ask if the convenience offered justifies paying double for the same compute and storage. They are using Kubernetes to reclaim that margin.
While the cost advantage drives initial interest, organizations moving databases to Kubernetes gain additional benefits that are strategically more important:
Developer velocity. The same 2024 DoK Report identifies faster deployment times as a primary benefit that organizations realize from running data workloads on Kubernetes. When databases use the same declarative infrastructure, CI/CD pipelines, and operational tooling as applications, provisioning time drops from weeks to minutes. The operational boundary between stateless and stateful services disappears, and developers gain the self-service capabilities they already have for application deployment, eliminating the ticketing, waiting, and handoffs that slow every release cycle.
Governance and compliance control. Proprietary DBaaS platforms often restrict control over data residency, encryption, and audit logging. Kubernetes provides a vendor-neutral control plane for applying security policies, managing access, and ensuring observability across all workloads, whether running on AWS, Azure, Google Cloud, or on-premises infrastructure.
Cost pressure, operational maturity, and strategic necessity are converging. Organizations that move databases to Kubernetes gain immediate cost savings, faster deployment cycles, and architectural flexibility that proprietary platforms cannot match.