Question on new serverless pricing model

Hi folks, I’m super thrilled for the new serverless option. I do have a question on pricing though.

The reading units, do they mean the number of records returned or possible scanned? For instance if I run a query on a collection that may return 1k results, but I’m paging it and just return the first 20, do I get charged for 1k or 20 read units?

Google’s datastore has an option for a keys only query that is cheaper than the regular query, is there support for such pricing model too?

Hi @Vinicius_Carvalho ,

I believe its the amount of scanned bytes to be placed in memory.

So if the query require the instance to scan x bytes this will be taken into billing consideration.

At the moment since serverless is new its the only option.