Wednesday, February 22, 2017

Update: SAP BusinessObjects Semantic Layer Webinar

SAP's Pierpaolo Vezzosi joined us on February 16, 2017, to talk about what's new in the SAP BusinessObjects semantic layer. Back on February 2, it was SAP's Gregory Botticchio who joined us to talk about what's new in Web Intelligence. Taken together, these webinars give us a great view of the current and future states of the SAP BusinessObjects platform.

Pierpaolo's presentation featured a wealth of information, especially for people working with HANA and BW. He spoke specifically about:
  •  Linked Universes
  • Universes on BEx query
  • HANA

Below, I provide a fairly detailed look at the topics discussed, but it would definitely be worth your while to view the webinar on demand, as Pierpaolo provides live demonstrations on all of these topics, as well as a lively Q & A at the end.

Direct & Online Access to HANA

SP3 saw the addition of Direct access and Online access as methods to access HANA, allowing Web Intelligence reports to access HANA data without the need to build a universe between them, with best practices enabled.

Pierpaolo provided live demonstrations of Direct and Online access to HANA from Web Intelligence using the newer DHTML interface, because, as he reiterated, this is where most new development will occur, due to the impending deprecation of Java applets.

He discussed a number of access methods in detail, and when to use them:
  • HANA online - best for exploration. It is fast and best practices are enabled by default. By default, you can return only 5,000 rows, but the user can override this default. However, "good" reports in Web Intelligence don't return this many rows. The big constraint with online is that all calculations and filtering are pushed back to the HANA system, which means that some complex Web Intelligence calculation are not available in this mode.
  • HANA direct on relational sources - has all the functionality of Web Intelligence, including the complex calculations not available in online mode, and the best performance / cost ratio for reporting. The recommendation is that you start the design process in online mode to discover the data and the shape of the data, then move to HANA direct mode to make all Web Intelligence functionality available.
  • HANA direct on OLAP sources - if you need hierarchies. OLAP on HANA direct may be marginally slower, so if you don't need hierarchies, stick with HANA direct relational mode.
  •  HANA via a universe - if you need to customize your access to HANA. For example, if you want to rename objects and do not have write access to HANA views, you can use a universe to make access more friendly, or if you want to add a layer of security on top of HANA, but in general, SAP does not recommend using universes for HANA access. The default should be to use HANA views.

Authoring UNX Universes on BEx Queries

You can author a UNX universe on a BEx query as a means of improving user adoption. Users have welcomed the ability to access BW directly with Web Intelligence, but many have found BEx queries to be not business-user-friendly. Building a UNX universe on a BEx query simplifies the view of the BEx query for business users. Pierpaolo provides a detailed demonstration on how to author such universes.

Linked Universes

When linked universes were introduced in SAP BusinessObjects XI, their purpose was to make complex universe projects easier to manage. This idea returns in BI 4.2 SP3: You can construct UNX universes and allow objects in those universes to be inherited by a newer universe. Further, when you make changes to the original (or master) universes, the changes are propagated down to the derived universe.

Linked universes allow you to:
  • Deliver a modular approach to universe building in the Information Design Tool (IDT)
  • Insert the objects of a master UNX into a derived UNX universe, with automatic updates propagation to the derived universe

Use cases for this linked universes include:
  • Component Use Case- using a team approach in which each team supplies one or more components for the derived universe (e.g., customers, products, measures)
  • Kernel Use Case - building a large data foundation (e.g., the whole data warehouse) in the original universe, then showing subsets of that universe in derived universes (e.g., for Marketing, HR)
  • OEM Use Case - building a common universe, which is then extended in derived universes for various customers

Again, Pierpaolo provided a detailed demonstration on linking universes, and it is definitely worth your while to view the webinar on demand for these demonstations.

APOS Migrator for Web Intelligence

Allan Pym, APOS COO, also gave a brief description of the APOS Migrator for Web Intelligence, which enables:
  • Bulk conversion of universes from UNV format to UNX format
  • High-volume re-pointing of Web Intelligence reports from the old UNV to the new UNX

No comments:

Post a Comment