This week, when I was picking up my kids at school, I ran into one of the nVision developers who used to work for me (who's still working at Oracle). She's working on taking nVision to Fusion, which will use the Siebel Analytics platform.

When I first remembered that Oracle's acquisition of Siebel included Siebel Analytics, I told my co-workers that if Oracle was smart, they'd use that as the replacement for nVision (and metric calculations in EPM).

So, Why is it a good thing?
Believe it or not, we at PeopleSoft evaluated Siebel Analytics (although under a different name) to be the infrastructure for the next generation of nVision. At the time, it was a small company called nQuire. We put together a bunch of things we wanted them to prove they could do within a week against multiple PeopleSoft systems. At the time, Chris Heller and myself had a lot of discussions about the product's claimed capabilities... his quote was "Either they're crazy or they're geniuses". We found out that they were geniuses.

Unfortunately for PeopleSoft, Siebel beat us to the punch and acquired nQuire and re-branded the product. Now that Oracle has acquired both Siebel and PeopleSoft, it looks like that original vision may be realized for PeopleSoft customers.

So, what is Siebel Analytics?
Siebel Analytics is most of the functionality we planned to provide in reporting in PeopleTools 9.

• A data abstraction layer that allows users to work with meaningful objects for building reports
• A browser-based way of building both tabular and crosstab reports by business users
• A server-based quering engine that runs the reports and delivers results

In the published PeopleTools 9 plans, we had different names for the same functionality

• Data Objects (Data Abstraction Layer)
• nVision Studio (Browser-based way of building tabular and crosstab reports)
• nVision Engine (server-based querying and reporting engine)

Mark Rittman has some good images on his weblog (which is a must-read for anybody wanting to understand what's going on with Oracle and BI) that shows screenshots for building and managing the data abstraction layer and for building queries or reports.

Data Abstraction Layer

The data abstraction layer looks exactly as it did when we reviewed the product.

The far right pane contains the list physical objects that are the sources of data and metadata. These can be tables or files (and I'm assuming XML sources now). You can think of these as record definitions in PeopleSoft.

The middle pane allows you to map the physical objects to objects that have a more meaningful business names and structures. They abstract away joins and unions and other physical attributes you don't want to present. (a good example we had them prove in the PeopleSoft evaluation was that you could have a single representation for sales, where under the covers you were getting historical sales from a data warehouse and current sales from the CRM system). These are related to Data Objects in the PeopleTools 9 feature list.

The left pane contains the presentation layer for how you would want to present these objects to business analysts. For example, you might want to have different versions of an order, depending on the vertical you're using (where fields specific to a vertical are displayed and not others, and where the terminology presented to the user is targeted to that user type). In other words, you might have a different presentation object for Universities and Commercial in CRM, where in a university a customer is displayed as a contributor and in commercial, they're displayed as customer. These are related to Data Views in the PeopleTools 9 feature list.

Reporting

Siebel analytics provides a drag-and-drop interface for building reports in the browser using these objects. Although I believe that there is some opportunity for making the user interface more targeted to business analysts, they've done the big, hairy effort to pull together crosstab and tabular reporting into a single reporting tool. Here's another screenshot from Mark Rittman's weblog that shows some of the user interface.

As you can see, you have the objects from the presentation layer in the left pane, and you have the ability to lay them out in a report.

So, What's left to be done?

Good question. Here are the major tasks I see that need to be done to get nVision using this platform:

1. Map ledgers to the Siebel Analytics metadata. This should be relatively straightforward, but it is dependent on the fusion business unit/setid project as well as how chart of accounts configuration gets done in fusion.
2. Teach the Siebel Analytics platform how to read trees. Again, this is dependent on the fusion tree project. Fortunately, there are a lot of designs in place for how an engine would utilize different approaches for modeling trees. Another good thing is that the Oracle database has sql extensions for trees/hierarchies that make this much, much simpler than what is currently being done in nVision.
3. Build a robust excel user interface that leverages the calculation engine in Siebel Analytics. From what I know about WebADI, this infrastructure may do the trick, because it is build to allow web services to be used interactively to embed application functionality into to provide a user experience targeted to a business user.
4. Extend Siebel Analytics with output management functionality (this will probably involve work with concurrent manager, which is being managed by the person who used to own PeopleSoft's process scheduler and report manager).

Conclusion

Hopefully this makes most PeopleSoft customers more comfortable with the future of things. I've already known or suspected much of this information earlier, but was waiting for Oracle to provide enough information publicly for me to safely write this entry.

Labels: Fusion, nVision, Performance