[seanhunter]

I don't have personal knowledge, but my understanding is that timeseries data in the oil industry is insane scale. Think 10s or 100s of wells in a field, sensors at regular distances in every well each taking multiple readings (temperature, flow etc) on a sub-second cadence. And a big oil company will of course have many fields.

It's in the same league as timeseries data in financial markets, which I have more direct experience of. Financial timeseries data probably has more metadata (trade/order flags etc) but in terms of raw data it's similar.

[cheriot]

What the article is talking about is an order of magnitude larger. Think millions of data points per minute coming out of a medium sized data center. It gets harder when you look at the cardinality of the data (explained better by the article), which corresponds to the sized of the indexes.