I got a very surprising bill from GCP. I used bigquery to query the google public patent database[1]. I thought I was using enough predicates for it to effectiv

I got a very surprising bill from GCP. I used bigquery to query the google public patent database[1]. I thought I was using enough predicates for it to effectively be a point query (using patent ID, etc). Turns out each single query was a full table scan of the 4TB dataset and cost about $5 each! I luckily stopped after 4 queries once my curiosity was satisfied, but I could have easily spent hundreds of dollars. It wasn't until some time later I saw my bill and realized what happened. They did do a "one-time charge reversal" for me. I didn't really feel satisfied with that response. I wanted them to admit such datasets are a huge liability for users. 1. https://github.com/google/patents-public-data https://github.com/google/patents-public-data