Itzel Torres propose we take this option off the table, since it seems that most implementations are using either a middle view tier or an ETL to a data mart method.
This option will make it harder to maintain with every change to ODS 3.x introduced a lot of model updates that resulted in schema changes that would break with previous versions. In the Ed-Fi Dashboard project we had to break with backward compatibility because of these schema changes.
Self-Contained++:
Need to define how we will evaluate the Analytics Middle tier for this option.
Would like to get a quick overview of the Analytics Middle tier from Stephen.
Open question is if we aim for generalized views that could be used by most or more prescriptive views that offer pre-calculated values?
All External
Lot or examples of implementations taking this approach, would they benefit from a core set of standard views? Can we come up with a list of must-haves for the next generation reporting tool?
Data Warehouse/Lake
Would like to explore a phased approach where we start to build up de-normalized standard views that would support/make it easier to extract data to a data mart.
Data mart should be built with the idea of it moving to a data warehouse in the future. It should be designed/modeled as a module in a star schema DW so that it becomes additive and a progression and the work does not have to be re-done.