The whole WebOps and database team was still there late.
They were provisioning servers for a new data center in Texas. The next major database cutover was supposed to happen the following day.
That was too much.
I sent a note saying we would not do both. We would finish the provisioning work, get some rest, and pick a new date for the database change.
It was a small decision, but an important one.
Technical teams can get proud of pushing through. Sometimes that is necessary. A production incident does not care that everybody is tired. A planned migration does. If a change can move, there is no prize for asking an exhausted team to stack another risky operation on top of the first one.
The message came from my phone and did not try to make the delay sound strategic. The team was still working. That told me tomorrow was not the right day. I would send the new plan when we had one.
That was it.
I had plenty of moments at Shutterstock when I wanted us to move faster. This was not one of them. The infrastructure was getting larger, the company depended on it more, and the cost of a rushed mistake was growing.
Leadership sometimes looked like pushing. Other times it looked like noticing that everybody had already pushed enough.
We moved the cutover.
A week later, the database change was complete.
It happened without an outage, which made the extra wait very easy to defend.