At a recent Boulder BI Brain Trust, Aster Data Systems discussed their company and products. A BBBT blog summarized our open discussion. One item (under NDA) was about the Aster nCluster-Cloud Edition offering available through both Amazon Web Services and AppNexus. The announcement and blog by Shawn Kung was released this morning.

So what? I feel that this is important because nCluster is available as a virtualized on-demand service. Without buying, installing, staffing and maintaining a data center, you can utilize cutting-edge database technology for sophisticated data analytics. If your BI group is the DIY kind, you can push the cutting edge even further. I hope that Aster will be innovative in marketing their offer through low-cost trial programs to introduce in-Database MapReduce (MR) technology to IT professionals.

An equally important item is buried on the second page. Aster also announced nPath, which is an SQL/MR extension that enables discovery of relationships between rows in a data set. For you SQL folks out there, remember writing SQL statements to compare one row with another, using UNIONs or subqueries or other ugly stuff. Now think of the SQL for time-series analysis. Now think of doing that in a (relatively) simple SQL statement that is efficient and scalable. I am studying this blog by Steve Wooledge for the technical details.

As a side note, Curt Monash has been tracking the MR influence for some time. I commend his DBMS2 blog, especially this posting on why MR matters to data warehousing. 

As a second side note, I discovered an interesting MapReduce counter-argument. Michael Stonebraker argued with Curt Monash that standard SQL-92 with User-Defined Functions (UDF) can do everything claimed by MR and hence MR is not needed. Does the MapReduce emperor wear any clothes?  Tasso Argyros, CTO and Founder of Aster, blogged in August about how MR takes UDFs to the next level. To say that there is controversy in our fair database community over MR is an understatement, and it started over a year ago. I recommend that BI/DW folks take note and understand the issues.

At the Boulder BI Brain Trust, we had a discussion with Stephen Brobst of Teradata about the issues and architecture for Operational BI. It is a great discussion! See the complete blog here.

Tony Rodriguez, Chairman and CTO of x88 Software, gave Jos van Dongen of Tholis Consulting and Mark Madsen of Third Nature, and me a demo of their new Panorama value-oriented database engine and Pandora data discovery/profiling tool.

Impressive demo! This product takes slicing and dicing of massive amounts of data to a new level. It has nice features for surfacing the formats or templates of data values, like street addresses, zip codes, etc. It also can search phonetically and probabilistically, so you can find 'loose' matches on a search phrase. And so on...

I wonder, however, about who would buy this product... A data modeling or data quality specialist would find this tool very useful. It certainly fits the requirements for an exploratory data warehouse. But, duplicating the terabytes from an enterprise data warehouse into this engine would not be feasible. If you target smaller companies as the platform for their primary data warehouse, will these companies have the sophistication to realize the business value inherent in this tool?

Take a look. What is your opinion?

As an educator, I have admired people who are able to explain complex concepts in simple, easy-to-understand ways. I still remember my freshman year at Caltech in my Feynman Physics class, trying to learning some pretty complex stuff. The cliche around Caltech was "It is intuitively obvious to the most casual observer" and then you unloaded with a pile of ugly stuff. Richard Feynman was different. With a heavy dose of curiosity, he did make that ugly stuff intuitively obvious.

I ran across another great example of explaining the complex simply. Take a look at this 3-minute video on Social Media. It is about making ice cream in a small town. However, it explains a board range of Web 2.0 dynamics at the same time. Enjoy!

The dominant theme at IBM IOD was the Information Agenda (IA). Do you have an information agenda? …was the challenge to all attendees.

The first session of the Gold Consulting briefings was by Ambuj Goyal, GM of IBM Information Mgt Software, the blog of which is here. Through the week, I was struggling with this theme. How much substance does IA really contain? How will IA influence IBM’s strategies and practices?

Part of the answer lies in the corner of the exhibit hall. Did you see the IA exhibit? I would not, except that a colleague - Piyush Malik, an IBM GBS Consulting Executive serving as the Global Delivery Leader for IOD and Business Intelligence solutions - dragged me over to the booth. Having been involved with the design and staffing, Piyush gave me a personal in-depth tour, guiding me through the IA Experience that shows how this approach can transform information into a strategic asset.

The focus was on the activities of the Henson family - typical family with a teenage daughter Lily who is about to go off to college in Switzerland. The family is confronted with arranging bank financing and insurance before Lily's trip and purchasing the right in-style clothes for college. Unfortunately on the way to the airport, they were in a car accident. Lily is hospitalized, and her luggage is damaged. The family is urgently confronted with medical issues and insurance issues. Luckily, integrated medical systems exchange Lily's medical records between US and Swiss hospitals, and insurance data was instantly transferred to quickly repair the damages. The story ends happily. Lily graduates and becomes manager at IBM. :)

At each stage, the story describes the difficulties and satisfaction that the Hensons experience at the digital hands of complex information systems. Behaviors of today’s customers are complex and continuously evolving. Through approaches like IA, companies can identify the proper services for customers like the Henson family, integrate the relevant data across their data silos, and properly design their business processes to leverage that information. In particular, the linkages into the pie slice of IBM methodologies and products were explained. Remember that

The pitch. . . If your company is lacking in understanding customers, then shape up! Let IBM assist in developing an IA tailored to your unique requirements!

My take on this. . . It was a bit corny but definitely refreshing to see/understand specific ways that typical families are dependent on complex information systems. For me, it gave new meaning to CRM. In other words. . . Help me, just don't segment me!

The point is to understand all the ways that the Hensons are (could have been) served by intelligent systems that integrates data into a single version of business reality, enabling innovative business processes. Hmmmm Now I am sounding a bit corny!

[Blog stream from IBM IOD/Gold October 2008 is here]

I did not want to miss the talk by Tom Davenport, an old friend from my days in academia. But I had to suffer though another blast from the LA All Stars. Am I showing my age? Is IBM legally responsible for hearing damage to the folks in the VIP section? Video on how information should be chasing us, rather than the reverse. Not bad! Punch line is "We can do this now!" What is this?

Robert LeBlanc, GM of Software Sales, opens on “What’s Your Information Agenda?” We need to optimized enterprise performance. Whether you are CEO, CFO, LOB, CIO all have the responsibility to drive innovation. He Global enterprise CEO expects substantial change (83%) and change in their business model (98%). Technical talent will come from new emerging countries. Corporate social responsibility has a positive impact on society, by 68% of the CEOs.

Sidenote: A quote that stuck me was “We are surprised by the impact that just a few consumers can have with their blog entries.” by Ron Hurlbut, CEO of Attune Corp. Hmmmm Is anyone reading this stuff? If so, please post a comment!

Robert introduced Tom Davenport, Professor of IT & Management at Babson College, and his research studies on the interaction of information and decision making. His title is Linking Information and Decisions: The Unfinished Agenda. Tom noted that we have made fantastic progress in storing and managing data, but progress over 20 years on decision support has been lacking.

Tom (who likes pies) sliced the information pie: decisions, control communication, processes. Heading toward a planetary realignment via: application agenda, automating processes, creating data, key unit is the transaction. Types of decisions researched: supply chains, etc. The results from these studies have found: wide range of decision perspectives, factors of structure level versus human-contribution level, difficulties of achieving One Version of Truth.

Tom suggested a pyramid figure with top-down levels for: Automated Decision – Structured Human – Loosely Coupled. [I am not buying the framework; too much is all mixed up] From bottom-up, proper requires are: info infrastructure to linkage to DP/behaviors to tight process/system integration. From bottom-up, technologies relevant: DW/analysis tools to scorecards/dashboard to workflow/rules/scoring. IBM is developing or acquiring all these technologies. Four-step decision sequence (which I missed because of a text message). Striving toward a Single Version of the Truth (SVT) is very expensive to do; hence, we need to deal with multiple versions of the truth.

Next was a panel with Noel Garry, Irish Life & Permanent, and Larry Williams, JM Family Enterprises . Good sketches of business challenges and how their specific information infrastructures were being remolded to support better decisions in specific processes. Putting together all the information is a prerequisite to decision support.

Robert wrapped up the session. What a great era! Why IBM? Answer offered: Global IT leader, technology innovation, proven solutions, industry focus.

My take on this. . . I was expecting a deeper exploration into the linkage of information to decisions. In particular, we need better thinking about allocating human judgment (involvement) to business processes. Tom touches on this. I would asset that ALL automated business process MUST have include some degree of human judgment. Anyone agree?

[Blog stream from IBM IOD/Gold October 2008 is here]