The web servers were already in Boston.
Now we had to send the customers there.
Moving equipment was only one part of changing data centers. The bigger moment came when real production traffic started using the new setup. A rack could look healthy for days without proving what it would do under the full site.
Our note to the team was short: the web-server move had worked, so we were beginning the process of moving front-end traffic from New York to Boston.
“Beginning the process” carried a lot of work.
The safest migrations were staged. Move a portion. Watch it. Check the systems from the customer side, not only the internal dashboards. Increase the share when the new path behaves. Keep a way back if it does not.
Shutterstock could not close for the move. Contributors were uploading images. Customers were searching and downloading them. The site had to keep running while its center of gravity changed underneath it.
This was a normal part of operating a company that had outgrown its earlier infrastructure. Growth created the need for more capacity and better redundancy. It also made every migration more consequential because more people were using the service while we did it.
The work crossed teams. Network, WebOps, databases, application engineering, and the people watching the business all needed to see the same thing at roughly the same time.
There was no clever line in the email. No victory lap either. The servers had moved. Traffic was next.
Then we would watch.
The site stayed live, which was the only version of the story customers needed to know.