Welcome to Part 3 – the finale – of the blog series “EDMCS and Data Governance!”
Part 1 provides an introduction and primer for data governance workflows in Enterprise Data Management Cloud Service (EDMCS) which was introduced in the 19.02 release.
Part 2 discusses Workflow Stages in greater detail and dives into the brains of EDMCS workflows – the Approval Policy. Approval policies at different levels of the data chain are explained, and we conclude by building a sample workflow at the dimension level.
In Part 3, I’ll attempt to tie a bow around everything and offer some parting thoughts.
As I continue to explore and learn about collaborative workflows in EDMCS, these are the key points that come to mind:
- Emphasize the Fundamentals – No matter what tool you are using, People and Process are extremely important in any data governance solution along with strong executive sponsorship and robust change management.
- Build the Foundation – get the client comfortable with the tool and content before you introduce workflows. A strong foundation (your applications, dimensions, views, and viewpoints) is needed before you start the plumbing and wiring (workflows).
- Brush up on Security – I haven’t discussed security extensively in this blog series, but the Oracle EDMCS User Guide does a nice job describing security requirements for assigning and approving workflow requests. Note that security enhancements have been introduced along with workflows. A new “Submitter” permission is now available to go along with Owner, Data Manager, and Browser. And permissions can be assigned at the Application, Dimension, Hierarchy Set, and Node Type levels.
- Ponder the Approval Policy – this is the most interesting one to me. As we discussed in Part 2, approval policies can be defined at 4 points in the data chain (see Figure 1). With the inheritance and inter-dependencies of approval policies across the data chain along with the actions each policy can govern, it is critical to efficiently design your approval policies up front.
o For example:
- Suppose your client requires a final “audit” type of approval across the board for any type of request for any dimension. Or they always a require an upfront “gatekeeper” type of approval to make sure the request is justified and complete before it continues down the approval chain. These would be good candidates for an approval policy at the Application level. And it would avoid having to define duplicative approval policies at lower levels in the data chain.
- Will your application contain dimensions that do not need data governance workflows? Then Application level approval policies should be avoided.
- Say you want to limit and govern the actions of a specific group so it can only work with existing nodes (insert, remove, update). An approval policy at the Hierarchy Set level is probably best.
o Overall, I believe approval policies at the dimension level are a good place to start. Then as the workflows evolve and requirements become more clear, you can determine if there are common factors across all dimension approval policies that can be consolidated at a higher level (Application level approval policy), or if there are specific subsets of actions that need to be broken out to a lower level (Node Type or Hierarchy Set level approval policy).
o All of which brings up another interesting point: effective approval policy design directly ties into effective viewpoint design. Think about it – you can define the set of Allowed Actions (Add, Insert, Move, etc.) at a Viewpoint level. Which means what? Special-purpose maintenance views are likely required to support certain approval policies, especially those at the Node Type or Hierarchy Set levels.
Figure 1 – Approval Policies and Data Chain
How do EDMCS Workflows Compare with DRM/DRG?
I was reluctant to include this section at first because in general, I don’t like comparing Data Relationship Manager (DRM) and EDMCS. Yes, they are both master data management tools and yes, they do share some common concepts and terminology. But overall, the two products are so different in terms of philosophy, deployment design, and underlying architecture that I think comparing the products is often less than helpful.
However, with data governance and collaborative workflows, I feel there is enough commonality that it is worth highlighting a few items. So here goes:
|User Interface (UI)||
|Separation of Duties||
I’m curious if Oracle will introduce a form-based UI for workflows. Part of me would very much like to see that so that you can present a clean user interface to the approvers, hide unnecessary details, and display special instructions and messages, but part of me does not. One of my favorite features of EDMCS is the visual highlighting of pending request changes and the “shopping cart” of request items that are displayed prior to submitting a request. I would hate to lose that by going with a forms-based workflow UI, but perhaps there is a solution that combines the best of both worlds.
Well that’s it, an initial look at workflows and approval policies in EDMCS. I’m excited to see how this functionality evolves and expands over time. Talk to you next time!
And don’t forget to follow me on Twitter (@kblackEPM) and check out these links for more information: