Specialized IT Services focused on Data Management | Speak with Us 877-634-9222
Art of BI: Web Catalog Statistics – New SampleApp – Quick Peek
If you’ve been looking at Oracle BI’s progress over the last few years, you know that the Oracle BI development team has been making a concerted effort to provide guidance about the Oracle BI platform, its integration capabilities with other software, and some best practices via the Oracle BI Sample App. Oracle BI 11g brought a very comprehensive new look and feel to the SampleApp endeavor along with a packaged development virtual image for those previously not so inclined to develop their own environments.
The latest SampleApp VirtualBox virtual machine image is due for release very shortly and I was glad to get my hands on a beta version of what is aimed to be SampleApp v205. I’ll try to host a quick series of post on what I think are the most powerful features, not only of the compiled SampleApp distribution but also of Oracle BI 220.127.116.11 (and patch releases for that minor version) over the next few weeks.
I’d first like to highlight a solution introduced in the SampleApp that many organizations have been yearning to have at their control, Web Catalog Statistics. This is a rather vetted solution developed by the Oracle BI development team, complete with PL/SQL ETL processes, a set of database tables added to the BIPLATFORM schema for storage, and several dashboards front-ending the solution in a rather meaningful way.
So, how many times have you or a client needed to see what dashboards or reports were not being utilized in your Oracle BI implementation? Or, which presentation columns belong to which Answers reports? And, which of those Answers/Analysis reports/request are on which dashboards? You get the idea. The Web Statistics functionality now available thanks to the Oracle BI development team allows you to answer these questions. It also ties into the existing (now enhanced w/ 18.104.22.168) Usage Tracking system. So this not only is a statistics tool but also a impact lineage tool for high to low discovery. BICG has a tool called IMPACT which is the only deep-level impact analysis tool that I’ve seen for Oracle BI at since it tracks lineage all the way to the ETL mappings for OBIA implementations as well as some other bells and whistles. However, the new web statistics is a good addition to the Oracle BI toolset.
The SampleApp has a dashboard, WebCatalog Statistics which is fairly comprehensive of the solution. The dashboard tabs/pages range from Overview and Dormant Objects to Columns to Analysis. There are enough prompts to give a high reflection of the web catalog and its current inventory to usage state. It has some nice graphs, and out-of-the-box general analysis that get you started but once you look at the subject area you can understand that much more in the way of intuitive analysis and KPIs could be derived.
SampleApp V205 (beta) has a really nice model set in the RPD back-end. In regards to the Web Statistics, the model for the related subject area is just an extension of the Usage Tracking Business Model. In fact the tables used for Web Catalog Statistics are stored in the BIPLATFORM RCU generated repository schema. This makes modeling easy. Also, joined in to the sample usage tracking model is the new auditing functionality of BI Publisher. We cover how to achieve the BI Publisher auditing functionality in the Oracle BI 11g Book; keeping it all under Usage Tracking the Subject Area is just brilliant.
As with a lot of quick-win solutions using a native solution works great. So, I was actually delighted to see that only three scripts (sh / bat) exist to create the schema tables, extract/refresh the web catalog reports, and refresh to schema tables with metadata. The code uses PL/SQL and SQL*Loader with a version for both core OSs (Windows / *Nix). I’m going with the assumption that all code under the SampleApp is still open source so you could of course tweak it for your solution, or operationalize it as Oracle does with a lot of their usage of open source software (Apache to OHS – I still love that one).
This is definitely a winner. I was a little thrown off by the use of “Web Catalog Statistics” instead of what I understand as the go-forward naming convention, which would make it “Presentation Catalog Statistics”, but that’s just me. Again, it is no BICG IMPACT and when I first looked through the solution, it gave me chills reminding me of some of the initial work that went into development iterations prior to the release of MDSearch version 1.0. However, I believe that every Oracle BI implementation going forward should in fact incorporate this Web Statistics solution without a doubt. And, now that you can view the Physical SQL sent through the Oracle BI Server, now stored in the Usage Tracking table, S_NQ_ACCT, the new Usage Tracking system is going to save people time and aspirin.