When you talked about 150M queries per day; are those inserts or reads? That's about 1,700 per second, which to be honest doesn't seem like a lot for time series metrics ingest. I would expect a single node on most TSDBs to be at least ~100 times that performance.
Can you talk about the data you were ingesting? Was it numerics, text, or something else?
Hi there -- We are currently ingesting 150M/day -- though it's not evenly distributed -- probably peaks around 3,000qps
Agreed that it's not too much yet. I think the element which kills us is needing to do rollup tables, and then query against both the raw data and rollup data for our customer analytics dashboard.
The data is a combination of date/time, strings, and ints. 22 fields.
