By scale
For teams
Small groups shipping together. The constraint is rarely budget. It is that every hour spent on infrastructure is an hour not spent on the thing you are actually building.
The constraint
A small team's scarcest resource is attention, and infrastructure spends it in places nobody chose.
What goes wrong
- The rota nobody agreed to
- One person learns how the deploy works, so asking them becomes faster than writing it down. The team then plans releases around one person's calendar, and nobody decided that.
- Interrupts cost the afternoon
- An alert at three does not cost the twenty minutes it takes to clear. It costs the rest of the day, because the work it interrupted has to be loaded back into somebody's head.
- Tools added, never removed
- Each service was adopted to solve one afternoon's problem and none was dropped. Maintenance is small per tool and large in total, and no single item is ever worth the meeting it would take to remove it.
What changes
What stops
Someone checks capacity before a launch, and again before they take a holiday.
What replaces it
Instances follow traffic on their own, and there is no rota to hand over.
What stops
A release is a scheduled event with a person watching the graphs.
What replaces it
A push to the default branch releases, and the version it replaced stays addressable.
What stops
Patches, base images and certificate expiry live in a shared calendar.
What replaces it
The runtime, the operating system and the certificates are maintained underneath you.
What stops
A new engineer spends their first week learning how to get code out.
What replaces it
Their first week is spent on the product, because getting code out is one command.
Questions we get asked
- We are six people. Are we too small for this?
- Six is the size where the overhead is most visible, because a week of maintenance is a sixth of the team's month. The size that does not fit is the team with nothing in production yet: there is no work to take over, and the platform is a decision you can make later with better information.
- A small machine costs less than this. Why not keep it?
- On the invoice it often does, and if your traffic is flat and someone on the team enjoys keeping it patched, keep it. The comparison changes when you price the hours at what you pay for them, and again the first time the person who knows the machine is on holiday.
- What happens when we grow, or when we want to leave?
- The application is ordinary code and moves with you. What stays behind is the deploy configuration, which is small, and the work you stopped doing, which is not: budget for rebuilding the parts you let go of rather than discovering them during the move.
What fits
Tell us what you are running and how many people are on it. If the honest answer is that you do not need us yet, that is the answer you will get.
Talk to us