ODS or EDW

I have a client that wants to do detail reporting from an EDW

The reporting information will include a lot of degenerate data such as notes, phone number, status and other header information.

The client would like the reporting area to be updated throughout the day so reports can be accurate. What an ODS is hard to determine since there is a lot
of different ideas. I would suggest to the client to use an ODS that gets updated throughout the day and at the end of the day the ODS will update the EDW.

The ODS design would be a combination of a star schema with certain tables being used for reporting. Such as the purchase order header where the needed information would be stored
and not be sent to the EDW.

Fact_ Purchase_order ---- POHeader

Dim_Vendor -- ---- Dim_Product

Any ideas if this is proper or not?

An ODS is typically a normalized model rather than a dimensional model because it is more suitable for the types of queries performed against an ODS.

Operational queries tend to be very specific, looking at individual activities for specific entities, rather than broad-stroke analysis of large amounts of data typical of strategic and some tactical queries a dimensional data warehouse is designed to support.

So, an ODS is usually a separate database. One thing you should consider is perform change detection in the ODS so it can serve as a delta feed into the data warehouse.

What you are calling ODS is actually a near real time data warehouse.

No.  "I have a client that wants to do detail reporting" sounds like the type of requirement an ODS typically supports.  It is the "from an EDW" part that is uncertain.

From an architecture standpoint, an ODS never comes from an EDW.  However it is common for an ODS to feed an EDW.  This is particularly true if your EDW is dimensional rather than normalized.  An ODS is typically normalized and could simply be a replication of operational database tables.

An ODS tends to be focused in scope to accomplish a specific purpose rather than be an open-ended, ad-hoc analytic environment.

If your EDW is normalized and carries all the necessary operational information, you could simply throw an interface over the EDW to support the need, but it is not an ODS. The term implies a separate data store.