A forray into Lumira SP12 – new features and findings
With Lumira desktop SP12 released last week, I wanted to take this opportunity to introduce a few of the new features!
Firstly, I recorded some videos to cover the 3 main enhancements:
- New HTML 5 User-Interface ‘look-and-feel’:
- “Compose Storyboards” – create interactive, composite views of visualizations:
- Data acquisition using BW-SQL .UNX universes is now supported:
I must also mention these incremental updates:
- Improved in-product registration.
- BIP 4.1 support.
- HANA SP06 support.
I thought you’d be interested to know some of the more techy stuff I discovered recently:
1. Whilst testing, I noticed 2 additional 32-bit Lumira process showing up in the Task Manager, and I was told that this SP12 release now uses a revised ‘Chromium’ multi-process architecture. With the interface now rendering in HTML, this task has been decoupled from the application’s JVM ‘back-end’ process. There are always 2 of these processes: one central/global process, one for the HTML page itself. More info here: http://www.chromium.org/developers/design-documents/multi-process-architecture
2. I was stress-testing the BW-SQL pretty heavily and made some interesting findings. Firstly, I had recently upgraded my BI Platform from 4.0 SP06 to BI 4.1 SP01 and was hitting errors like “getSessionInfosEx – could not load the selected universe” (and the ‘csEX’ error below) when trying to acquire from BI Universes:
As it transpired, my problem was related to Adaptive Processing Server (APS) configuration on the BI Platform. After updating to 4.1+ , my old 4.0 APS configuration wasn’t working with Lumira (although reporting content was fine from BILaunchPad). I suspected a problem with the Adaptive Connectivity Service. To solve the issue, I ran the CMC’s new “System Configuration Wizard” (thus recreating all APSs) and then my acquisition worked! 🙂
So now i could connect to BW-SQL universes 🙂 My next issue was when consuming very large infocubes.. I had expected some slowness as my cube had several million rows in its fact table, but eventually I hit an error:
The solution was to increase some specific parameters on the Web Intelligence Processing Server (default shown in black outline):
The crucial parameter here is the “Maximum Character Stream Size (MB)” setting. This needs to be upped from 15 to anything between 40-100Mb (depending on the # of rows being fetched). Although this hasn’t been absolutely proven empirically, the “ball-park figures” i’m communicating are in the range of:
- 100 000 rows -> 20 MB
- 200 000 rows -> 40 MB
- 500 000 rows -> 100 MB
3. Unfortunately, there are a number of ‘release restrictions’ which need to be kept in mind. The article of interest can be found here www.service.sap.com/sap/support/notes/1911965
4. If connecting to BOE XI3.1, it has come to our attention that you can see MDX OLAP .UNV (BAPI access) listed in the universe browser – but this type of data access isn’t supported!! A bug has been logged (they should be hidden) . Many thanks to Bart Bart Verweijen for the feedback http://scn.sap.com/docs/DOC-33652
5. As being discussed here http://scn.sap.com/thread/3433313 for Predictive Analysis users, there will be no SP12 release (and probably no SP13). These applications have ‘de-couple’ temporarily, but will resume by SP14. However, this presents a dilemma for PA SP11 users who want to use Lumira SP12 – as there’s currently no officially supported way of having the 2 products installed side-by-side..
6. *update* With regards to this old article: 1782010 – SAP Visual Intelligence 1.0: How to analyze VI processing problems using the native Debugger utility
I realized last week that this tip no longer applies in SP12! The equivalent debugger/profiler is accessible in SP12 by double clicking on the ” – ” symbol in the status bar:
I hope this information is of use to you.
Please feel free to ask your questions here, i’ll be happy to respond .