We have c. 60 tables whose changes we would like to capture (CDC) using GCP's datastream and ingest into our data lake.
Are there any drawbacks to using Datastream? And should I set up one stream that ingests all tables? Or should I create a stream per group (or per single) table (in order to create a fail-safe that if a stream fails, it will be localized to a few/specific table)?
Thanks in advance.