A Deliberate Hybrid Cloud Strategy — Not All or Nothing
Problem
As Regal’s IT function matured, the natural pressure was to follow the industry default and move everything to the cloud — SaaS for everything, infrastructure-as-a-service for the rest. But a blanket cloud migration doesn’t fit every workload, especially for a $150M company where SaaS costs scale with headcount and usage in ways that don’t always track value delivered. At the same time, staying fully on-prem forever would leave the company carrying operational risk (power, cooling, hardware refresh cycles, DR) for workloads that are genuinely better run by someone else. Neither extreme was the right call — the decision needed to be made workload by workload.
Approach
I approached this as a hybrid strategy rather than a single migration project: critical, high-change-risk, or naturally cloud-native workloads (e.g. [fill in: WMS/Prophet 21 moving to Azure, phone system, etc.]) moved off-premises where the total cost and reliability tradeoff favored it. Meanwhile, I invested in a self-hosted foundation — a Packer → Terraform → Ansible pipeline building golden images on our own Hyper-V cluster, with k3s as the container orchestration layer — specifically so that high-value internal applications (starting with an internal contract/invoice tool) could be deployed quickly and cheaply in-house, without either paying SaaS margins for something simple or building bespoke infrastructure every time.
This gives the department two levers instead of one: SaaS/cloud for things that genuinely benefit from it, and a fast, repeatable internal deployment pipeline for everything else — so the default answer to “how do we run this new internal tool” isn’t automatically “buy another subscription.”
Outcome
The result is a company that isn’t locked into any single cloud vendor or SaaS provider by default, and a two-person team that can stand up a new internal application on existing infrastructure in a fraction of the time (and cost) a from-scratch buildout or a new SaaS contract would take. The IaC foundation also means the on-prem side of the hybrid model isn’t a maintenance burden the same way ad-hoc on-prem hosting is elsewhere — it’s repeatable, documented, and version-controlled.
[Fill in: first app live on the k3s platform, any concrete cost avoided by not going full-SaaS, or deployment time for a new internal app once the pipeline was in place]