I'm Pratyush Shivam, CTO of Konfidio in Berlin. A line I repeat to every team I work with: designing a network is one challenge; running it reliably every day is another.
Where the attention goes
Design work is visible. There are diagrams, reviews, a launch and a demo. Operations work is mostly invisible when it goes well, which is exactly why it gets underfunded.
What reliability takes
- Monitoring that speaks first. You should know something is wrong before a customer tells you. If your first alert is an email from a user, the monitoring isn't finished.
- Upgrades without downtime. Plan them, rehearse them in a copy of production, and keep a way back.
- Runbooks written in calm weather. The worst time to work out how to restore a service is during the incident.
- Post-mortems that fix causes. Blameless, written down, and followed by a change that makes the same failure less likely.
Why it matters more for shared systems
When several companies rely on the same network, an outage isn't one company's problem. Every participant feels it at once. That raises the bar: shared infrastructure has to be boring in the best sense of the word.
The takeaway
Build for the ordinary Tuesday, not just the launch. The systems people trust are the ones that quietly work on the days nobody is watching.
More of my writing is at pratyushshivam.com, and my paper on what permissioned ledgers can and cannot prove is open access at doi.org/10.5281/zenodo.24438467.
Top comments (0)