I love postgres and my business relies on it. However, at the scale the author is talking about (~100m rows), throwing all the data into BigQuery is very likely to be a better option for many people.
If rows are around 1kB, then full-dataset queries over 100m rows will cost less than $0.5 each on BQ -- less if only a few columns have to be scanned. Storage costs will be negligible and, unlike pg, setup time will be pretty much nil.
If rows are around 1kB, then full-dataset queries over 100m rows will cost less than $0.5 each on BQ -- less if only a few columns have to be scanned. Storage costs will be negligible and, unlike pg, setup time will be pretty much nil.