3 ms·
How do you find the querying costs? It seems like for many use cases, you'd have to pay quite a bit of attention to partitioning the database in a way that min
by antonvs 8y ago
How do you find the querying costs?
It seems like for many use cases, you'd have to pay quite a bit of attention to partitioning the database in a way that minimizes query costs.
- minimaxir 8y agoThe relatively new date-partitioning feature for tables has cut down query costs/time significantly.
- fhoffa 8y agoAnd with the just announced clustering now you get even better costs and performance. Disclosure: I'm https://twitter.com/felipehoffa https://twitter.com/felipehoffa and I work for Google Cloud. And I'm really excited to reprocess all public tables into clustered ones.
- gipp 8y agoCost (or rather bytes scanned) is sent back from the API on any queries or dry run/validation calls. There's a common chrome extension to put it in the legacy ui, and I believe the beta ui has it baked in (as part of validation)
- djhworld 8y agoFor this trial we are transferring some data from AWS, which is already partitioned in S3 by quite a few dimensions, including DAY. As BQ only supports DAY partitioning, I had to create multiple tables that represent the other dimensions, but it seems to work OK, if a little clunky when you have to use wildcards/_TABLE_SUFFIX