问题
Can I denormalize (by joining) multiple large tables in bigquery?
Table1 is 400M rows Table2 is 2M rows Table3 is 800K rows
If not, do I have to do it in my relational database before I upload? That would be a difficult solution.
Should I chunk the tables into smaller pieces and run iterations of joins over the tables so that it is always large to small? This would also be a difficult solution.
Thank you.
回答1:
BigQuery now supports "Big JOINs" which allows you skip the LIMIT's in your JOIN queries.
Docs here: https://developers.google.com/bigquery/docs/query-reference#joins
回答2:
Yes you can make new tables based on query results.
Try to look here:
https://developers.google.com/bigquery/docs/queries
and here:
https://developers.google.com/bigquery/docs/tables#addmoredata
来源：https://stackoverflow.com/questions/15287166/can-i-denormalize-multiple-large-tables-in-bigquery