Deploy daily and still pass the audit
Ship faster inside the controls
Sound familiar?
- A change takes a month because the change process runs on meetings.
- Risk and audit are negotiated by every team, every time.
- Cloud is approved in principle and blocked in practice.
- Your engineers want to work this way. Your organisation doesn't know how to let them.
What I do
I build the delivery platform as a product, with your development teams as its customers, and I get risk, security and audit to agree the controls once, inside the pipeline, so no team has to negotiate them again. Then I take it bank-wide: an academy that trains whole teams, a playbook that says what good looks like, and a community of practice that carries the change after the coaches move on. I've done this with Puppet, AWS, Azure and Google, and I've spoken about it at PuppetConf and in "Controlled Evolution with Puppet and AWS". Where the goal is to leave the data centres, I run the migration the way Deutsche Bank did: a dedicated cloud operations team, the easy-to-operate applications first, duplicates retired rather than moved, finance rebalancing allocations as applications move, and the hardware contracts reviewed so nobody is penalised for leaving. Six applications a day is the pace to plan for. The change management underneath all of it follows the diffusion of innovations: innovators and early adopters first, the rest when they can see it working.
What it has produced
Scotiabank Global Banking and Markets: the bank's first fully automated infrastructure platform and the first production application on public cloud at a Canadian bank. Infrastructure support effort down 90%. Deployment frequency up 640%, to 4,550 changes a week. Running in AWS's new Canadian region within 90 minutes, which AWS announced itself. → Scotiabank GBM platform
Then bank-wide: agile and DevOps across more than 10,000 engineers, a 465-person community of practice, and Wealth technology operating costs down 12%. → Scotiabank agile and DevOps adoption
Deutsche Bank: 25,000 technologists in 540 teams onto one toolchain, 10,000 of them in six months. → Deutsche Bank developer tooling
How we'd start
One platform, one division, twelve weeks: the pipeline with the controls in it, the first teams on it, and the numbers (deployment frequency, support effort, lead time) measured before and after. The academy and playbook follow once the numbers are in.