META system design round: Design backend of monitoring tracing System like Datadog/Signalfx/Dynatrace

A colleague of mine recently went for meta L-6. During the system round he has been given a problem statement to create the backend core of monitoring system like Cloudwatch, SignalFx, DataDog.

He strated off with Time Series database but faltered on when asked about ingestion of 10 million records/second scale.

After he came back we both discussed and summed up in two video

https://youtu.be/XKO9zLbbMfc?si=YdZnNw-5h7990e5b

https://youtu.be/KZSwYELGFqI?si=DbFTyjySdYkv82Ci

Do check it out. It might help.