This year was the first time I attended the RittmanMead BI Forum at Brighton (it has the annoying tendency to be on my wife's birthday, this year was no exception).
The first day Master Class on Big Data was very interesting. It was nice, up to date, 4 parts description of the modern Big Data world, by Lars George from Cloudera. I advise to have a look at his slides when they are available at RittmanMead site, after the Atlanta event (I hope).
It's funny but it seems to me, that most Big Data experts, don't "think BI" the way my colleagues and me do, despite the fact Big Data solutions mainly do data analyzing. It's the little things, the examples... Their background is usually software developing.
Later that day a great keynote by Matt Bedin from Oracle about BI Roadmap. The future seems to be very interesting, good looking and covered by (Oracle) clouds.
During the next 2 days there were 2 presentations about OBIEE in the Cloud.
There were many other good and interesting sessions.
Some of them were about Times Ten and Exalytics. The bottom line is: While using brute force approach of having all your data in memory might give you some benefit, you will have great results if you use your head (Emiel Van Bockel, CB Logistics presentation) and follow some well defined steps (Chris Jenkins, Oracle presentation).
They were 3 presentations that made me really happy I came to BI Forum all the way to Brighton:
1. Robin Moffatt, Rittman Mead, “No Silver Bullets : OBIEE Performance in the Real World” (you can see it here) had a great presentation about analyzing the OBIEE performance. Unfortunately the slides only cover part of what he said.
2. The great/funny/interesting/too short presentation by Christian Berg, “Neo’s voyage in OBIEE”. I'm sure that even people that have no knowledge of OBIEE, would have enjoyed it. Having almost 9 years of OBIEE experience, I was delighted. He talked about his investigation of the NSQ calls (NQSExecCalcScript, NQSQueryMetadataObjects...). He did a great investigation.
3. The presentation by Andrew Bond (Oracle) and Stewart Bryson (Rittman Mead) was interesting and funny. Each time I saw Andrew he has done good standup show with deep dive to Architecture. He talked about an updated version of the Enterprise Data Architecture that includes big data, but it was during Stewart Bryson's part that I had the religious moment of seeing something that was always there, in front of me, and I didn't notice.
During the forum there were few other interesting sessions about faster BI project development (Waterfall vs. Agile, having the same process picture... by Paul Oprea from Dufrain Consulting, by Michael Rainey from Rittman Mead) Stewart was the last to talk about the subject:
We all agree that long term waterfall attitude is great for creating lots of "CYA" documentation but takes too much time, just to be rejected by the user, in many cases. Stewart suggested to use OBIEE Metadata for fast development:
Use your raw data sources at physical layer. Do anything you can to create with it simple/complex objects in the Business Model Layer. Use them for rapid creation of results you can show the user. Don't mind performance. This can be done in hours / days instead of weeks / month. Now both of you see the same reports and dashboards. The objects in the Business Model Layer are the guidelines for you ETL.
I often do something similar, but never formalized this approach.
Some more references to the forum:
Great Confernce.
Boris -- glad you enjoyed my presentation, and sorry I didn't get chance to say hello in person.
ReplyDeleteThank You Robin,
DeleteGreat presentation and very justified title of best lecturer.
Good stuff dear. Windows Thin Client & Citrix Thin Client
ReplyDelete