DevOps and the Modern Data Experience — Permissionless Innovation
Tristan Handy connects DevOps principles to the data stack, arguing that the project of the next decade is not technical but experiential: creating a system of knowledge creation and curation accessible to all knowledge workers.
Key mental models
Permissionless innovation
Hashicorp built Terraform providers without AWS’s permission. Great APIs and standards enable users to build abstraction layers on top of products, finding new and exciting uses. Permissionless innovation is fundamentally unlocked by APIs and standards.
The modern data experience is “purple”
Data stacks have a history of building walls to throw things over: engineers throw pipelines at analysts, BI developers throw reports at stakeholders, analysts throw results at anyone who will listen. The modern data experience breaks down walls through cross-functional collaboration.
Pets vs. cattle
DevOps moves systems from “pets” (hand-fed, irreplaceable) to “cattle” (numbered, replaceable). This enables engineers to hit the delete key. Applied to data: dashboards and models should be disposable and recreatable, not precious artifacts.
The data literacy framing
“Imagine your company as a society where only half the population can read, one tenth can write, half a dozen languages are spoken, and most books contain outdated information.” The next decade’s project: everyone must be able to read and write data.
Kill your darlings
Agile encourages editing — new information makes the final outcome stronger. Apply this to data projects: don’t cling to initial assumptions.
Connects to data team operations, analytics engineering, tools and infrastructure.
Open questions
- What does “the entire stack is scriptable” unlock for data teams specifically?
- How do you move a data team from “pets” to “cattle” when stakeholders treat dashboards as precious?