DEV Community

Discussion on: Choosing Your Data Warehouse

Collapse
sheyd profile image
Sena Heydari Author

Evan,

I might be misunderstanding your question, but let me provide 2 answers:

  1. When you mention RDS, I'm guessing you mean MySQL, since RDS has PostgreSQL, MySQL, MariaDB, etc. as products you can use. For MySQL, we don't have consistency issues for the ETL work that writes to the master node, since MySQL (in most cases) errors out if a transaction can't be consistently completed.
  2. If you're referring to the eventual consistency of slave DB replication which powers our analytics and dashboards, we've run into a few issues, but we have a few safeguards against it:
  • Our data team knows that the replication is not always real-time or close to real-time, especially when large amounts of records ingestion or processing is taking place.
  • As long as the replication delay falls within our data delivery SLA i.e. our dashboards are no more than X hours out of date, we don't escalate to notifying the entire org.
  • If our data delivery SLA is exceeded, we notify staff and make sure to notify them again when things are back to normal.

Granted, these are operational safeguards, and there could be scenarios when these might fail us. But this is our first pass. Let me know if I didn't answer your question fully, and I'm happy to jump back on this thread!

Collapse
evantbyrne_38 profile image
evantbyrne_38

Sena,

I was referring to the eventual consistency on the slave DB replication. This clears things up. Thanks for the detailed response!

– Evan