Data Modeling for Integration of NoSQL with an RDBMS Data Warehouse
This post for data modelers, like a good portion of my online content, is in the form of published slides. To view them, click here. If you like them, feel free to subscribe here for the occasional new entry.
One thought on “Data Modeling for Integration of NoSQL with an RDBMS Data Warehouse”
Immediately below, I’m pasting a reader’s emailed question here to move the dialogue back into the blog.
I was wondering if I could ask your opinion on an integration use case
that I’m currently analysing as it also seems to be an area of interest
I’m integrating (ignore definition for now) a traditional Kimball Star
schema with a NoSQL (Columnar) database for the purpose of
reporting-only. In other words, at this stage of design, the
architecture only requires the source data to be “available” within the
NoSQL repository so that the concomitant reporting tools can be
I was curious if you had any thoughts about this use case as it
effectively fails to exploit the NoSQL paradigm and merely mandates
integration for the purpose of displaying structured data via a
pre-designed dashboard that can only serve data from the NoSQL repo.
My thinking is to merely flatten the Star schema via an SQL query and
then to send this as a single dataset over to the NoSQL repo ready for
reporting. My interest here is really how we define integration between
these systems in this scenario. For this use case, it seems to be no
more than the lifting and shifting of a dataset to a new repository
whereby it will be structured differently (i.e. columnar).
Any thoughts you have on this use case would be appreciated. I also
enjoyed your presentation which I felt captured the essence of a problem
not widely understood at this present time.