Performance to do full table scan using Google Spanner

Google Cloud Database
1

I know that Spanner is for OLTP use case but we are the real time OLAP use cases that needs full join support with large dataset.

Spanner’s perforement on full table scan is a concern to us, given this benchmarking Google Cloud Spanner: The good, the bad and the ugly | Lightspeed POS

Is it an anti-pattern that we should not do (full table scan) or something we can do but not perfect? Do we have latest benchmark on this with table like 1B records and tens of columns? The one from 3 years ago said it could take 1 minute with a 10M record table.

Thanks!

2

Hi @liliangc The article you referenced is quite old, as you noted. As you might imagine, we make continual progress on performance primitives like scan speed and being smarter about data access/retrieval like reducing the types of queries that will perform full table scans. We don’t have any public benchmarks at this time but if you can share details about your scenario (including schema and query patterns), we can help you with optimizations. Feel free to DM me if you don’t want to share details publicly.