Alow multiple SQL DataWarehouse instances to access the same data (.mdf files)
This would allow an enterprise to have a Production Data Warehouse with SLAs in place while allowing a data scientist/Business user/Subject matter expert to do analytic and discovery work without impacting SLAs and without moving data (which usually is pretty large).
At the same time it would allow for metered consumption and chargeback towards the subtenants that are conducting analytic discovery work.
The functionality would be similar to what can already be done today with HDInsight and Azure Blob. Data in Azure blob can be independently accessed by multiple HDInsight clusters.
Thanks for the suggestion. We are looking into this scenario for a future release. 7557959
Frederic Forest commented
This is currently the main killer feature from competitor Snowflake (that and near-instant scale up/down).
This feature allows us to have independant compute instances per project, all taping in the common data warehouse underlying data. It makes a WORLD of difference in DWH management
this could also be used to allow more concurrent users.
it could be handled in read only (dirty read) mode
Mark Webb commented
I think this is a very high priority with a primary objective being to eliminate the rampant data replication of multiple data domains. Also fosters a single "source system of reporting/analytics record" and allows for application of data element naming standards (with business intuitiveness) not currently enforced at the ERP level.