Infrastructure

Cloud Architecture & DevOps Consulting

Where I can help

• Deployment pipeline design and build automation for teams releasing manually or unreliably
• Database migration planning, including cloud-hosted MSSQL upgrades and moves
• Containerisation and environment consistency using Docker
• Reviewing an existing cloud setup for reliability and operational cost
• Advising on architecture for teams without a dedicated infrastructure engineer

Migrations, specifically

Database migration is the part of cloud work where inexperience is most expensive, because the failure is often discovered days later in the form of data that quietly did not arrive.

Managing cloud-hosted database environments for US medical practices meant treating every migration as something that had to be reversible. The parts that matter are rarely the copy itself:

• Establishing a verifiable record count and checksum on both sides, rather than trusting the tool's success message
• A rollback path that has actually been tested, not one that exists only in the plan document
• Understanding which application behaviour changes under the new version before cutover, not after
• Planning the cutover window around real usage patterns rather than nominal business hours
• Deciding in advance what constitutes a failed migration, so the decision to roll back is not made under pressure

A note on honest scoping

If you need deep, specialised platform engineering — large-scale Kubernetes operations, or a dedicated SRE practice — you should hire a specialist in that. I am most useful to teams who need pragmatic infrastructure judgement and hands-on migration work, not to organisations already running a mature platform team.

Equally, if your deployment process is currently one person running commands from a laptop, the highest-value work is usually not a sophisticated platform. It is making the existing process repeatable, documented and recoverable. That is unglamorous and tends to matter more than the architecture diagram.

Being clear about that boundary tends to save everybody time.

Common questions