Elasticube Build Behaviour - Upsert

[Camantea]

Currently, the "Upsert" build behavior is only available for Build to destination. It would be great if we were able to utilize this feature in our Elasticube model where we are not building to destination. We have numerous tables containing created date as well as updated date that are attached to an id field but are unable to use the updated date field since this will bring in duplicate (or more) rows per id which we can not have. Ranking duplicates and filtering out is not a sustainable with the size of our datasets.

The ability to look at the id and updated field would solve this issue without the need for a complex workaround in our data source (Big Query).

Thank you!

[Jango]

Thank you for your feedback Camantea, many customers are using this workaround.
Having said that, we are aware it is not a scalable solution and we are working on another alternative as we speak. The project is in research phase with an improved Elasticube version. In case you are interested to take part in a beta program please approach your CSM for them to connect you with product and the program.

[Jango]

Hello @Camantea,

Thank you for sharing your idea in the Sisense Community. We value your input and think this is a great idea! 

This is a common ask, but due to product constraints, it will require further investment. So, in order to prioritize our development efforts, we want to gauge community interest. We kindly request you to encourage other users to vote and comment on your suggestion. The more engagement your request receives, the higher the likelihood it has of being considered for implementation.

We appreciate your efforts!

Best, 

Nir Vaknin

[AndrewLoomis]

Hi @Camantea ,

I see you are using BigQuery as your data source.  If you could build the tables directly into BQ and then connect live would that solve your problem? Your input is greatly appreciated!

[Camantea]

@AndrewLoomis thanks for the suggestion / thoughts on this!

Yes, we are already using live models in some instances but can not utilize live all the time as a workaround. Would not want to query BQ constantly throughout the day from a live model (driving costs up) and load times are significantly decreased on live vs. cubes - not to mention you can only utilize one data source in a live model and can not build custom columns. 

This is why a permanent solution vs. numerous workarounds would be very beneficial.