 Julia's interview. Take one. The monitoring at the time could not keep up with instances coming and going. You have so many more machines, you have so many moving bits and pieces. You have to find out that it is broken. Now you have to find out what is broken. And yeah, there was just no chance. You have to think about buffering, greatly limiting, and lots of complex stuff that ideally you don't need to worry about. Teams are working on small pieces without talking to other teams, so you really need observability at that time. People had started catching on to metrics as a base of monitoring. There was no system at this time in the open-source space, which was either too simple and not powerful enough, too powerful, but not scalable or too complicated for a cloud native. A approach that we got used to at Google was really something we wanted to see in the open-source world. They had to scramble and really figure it out. As kind of a hobby project initially, we just sat down in our free time and started building what eventually became Prometheus.