My team uses AWS for all of our servers, CDN, SQL and cache boxes but use BQ for all of our telemetry data - about 20 billion metrics a month.
Not having to manage a redshift cluster and just let BQ do all the work for us is worth the query time; even then, most of our complex queries run sub 30 seconds over hundreds of GB. If your data can be queried using a date-range then use date-partitioned tables. There is a 1000 table query limit so there are some pit-falls but BQ makes it easy for you to aggregate data into weekly and monthly tables to query against.
Not having to manage a redshift cluster and just let BQ do all the work for us is worth the query time; even then, most of our complex queries run sub 30 seconds over hundreds of GB. If your data can be queried using a date-range then use date-partitioned tables. There is a 1000 table query limit so there are some pit-falls but BQ makes it easy for you to aggregate data into weekly and monthly tables to query against.
Disclosures:
1. I don't work at either.
2. My employer, however, partners with both.