Connector Improvement: Fivetran is changing data types between same source & target for no reason
AnsweredReference case #197836
Fivetran is converting ARRAY column in source to VARCHAR column in target. Both source and target systems are snowflake here so ideally everyone would expect a data type in same source system to remain same in the same target system which is snowflake here.
Fivetran should not change the data type in between as it's a very bad experience for the end users.
Please let me know if you need more details.
-
Official comment
Hi Sudhir,
This connector improvement is not currently on our roadmap. Today, we convert data types to our platform’s standard types to ensure compatibility with the selected destination—especially when it differs from the source. We do understand that in your case, the source and destination are the same, which makes this behavior less ideal.
Feature prioritization is guided by community interest and upvotes. Each upvote on this request strengthens the case for consideration. Additionally, sharing more details about your specific use case—and the impact of converting
ARRAYtoVARCHARin your environment—would help us better assess the need and justify future investment.Best regards,
-
Hi Egidio,
If downstream logic is written considering source column is ARRAY then they would expect same datatype in target specially when source and destination both systems are Snowflake databases.
Ex. Business wrote their logic using some native array functions in snowflake but it failed as Fivetran convertedARRAYtoVARCHAR on it's own. I could have understood this if source and destination systems are different and fivetran uses native datatypes to manage compatibility between systems but for same type of source & destinations, This is a surprise to end users and also a bad user experience. Even old database replicators like Oracle GoldenGate used to take care of this seamlessly and I believe Fivetran being a modern replicator platform should have this feature too. Thank you.
Please sign in to leave a comment.
Comments
2 comments