CODEWITHSUNDEEP
databasejoingoogle-bigquerybigdatadenormalization
IAmHereToParticipate
IAmHereToParticipate

Reputation: 31

Is denormalization more useful in big query?

Is denormalization more useful in BigQuery given that BigQuery does not support indexing? In a traditional RDBMS we can index columns, or use foreign keys, to speed up JOINs, thus making denormalization not worthwhile in most use cases. However, given that BigQuery does not support indexing, I am assuming BigQuery would require a flattened model in order to avoid the additional scan that would of come with a JOIN clause. Is my thinking correct here?

Follow up question: Are denormalized tables recommended in order to save on compute or is using one or two JOINs negligible even with large datasets?

I tried to test performance differences in big query with both a normalized and denormalized table but was unsuccessful in getting meaningful results with the minimal amount of compute that I was working with.

Upvotes: 3

Views: 451

Answers (0)

Related Questions