Open Source Business Intelligence: Seth Grimes
Seth Grimes, a consultant specialized in large-analytic computing system, who in 2003 was engaged by Pentaho for market-positioning, few days ago stepped by Rome to present a course, entitled Open Source for the Enterprise.
Seth a couple of weeks in advance wrote me an email to get in touch, and last sunday we meet to talk about the open source market, business intelligence and related stuff, as follows.
Seth, how do you spend your day?
Like you and many others I know, I have a fork in several pies. I spend about 40% of my time doing hands-on work relating to management, analysis, and dissemination of government and marketing statistics. Another 40% of my time I spend on IT strategy consulting, focusing on business intelligence and text-analytics technologies. The remaining 20% of my work is writing, presenting, and teaching — same topics, a variety of audiences — my Rome class on Open Source for the Enterprise, the Amsterdam Text Analytics Summit the week before, and so on.
Are you deeply into Open Source?
I’m not an open source specialist. I work at the applications layer, and I’ve found over the years that open source is often the best option measured by a combination of capabilities, ease of introduction and use, and cost. I’ve been using Python since the mid-’90s and Linux, MySQL, PHP almost as long. Over the years, I have become a fan of Apache, Mozilla, OpenOffice, and a slew of other Web and end-users tools.
I suppose that my predisposition to open-source was helped along by my use of the Internet. Nothing unique there: the Net in earlier days was about connecting and sharing, a natural for those who are community minded. I think I first used Network News (Usenet) and sent my first international e-mail over Bitnet in 1984, and starting in the late ’80s, I was an Internet (and then Web) evangelist at a series of organizations where I was employed. But I actually trace my involvement in OSS — in commercial OSS much earlier.
I first learned to program in high school in the mid-’70s. We wrote Basic code that was interpreted, not compiled, so the source code was exposed. Time-sharing users had access to program libraries: utilities, applications, games, etc. I spent many hours playing a Star Trek inspired space wars game — this was dial-in on a 110 baud/10 CPS teletype with an acoustic-coupler modem — and I coded a slew of enhancements and improvements. The commercial part: my friend Mitch and I went to Star Trek conventions in New York in ’75 and ’76, and the second year there I brought listings of my modified code and even a couple of copies punched out on paper tape, and I sold a couple to one of the exhibitors there. I think I got $10 each.
Tell us something about Pentaho, and Open Source BI.
Given my BI interest, I first surveyed open-source options back in 2002. I got a chance to use some of the software for real work starting in late 2004. I was hired to introduce BI at a Washington DC membership association, which had very limited in-house IT skills because all their applications — management of membership, bookstore and software sales, meetings, knowledge communities, continuing education — was hosted. In keeping with their modus operandi, the organization budgeted lots of money for consulting and nothing for software. So I set them up with MySQL, Mondrian OLAP, and JPivot for JSP interfaces, and I did my data work with Python. In retrospect, we should have spent more effort building a BI culture, figuring out how to incorporate analytics in everyday operations. The system funtioned well enough technically; acceptance obstacles had nothing to do with open/closed source software origins.
That said, it OSBI of the era — and I think this is still largely true — was technology for Java developers. It took a lot of work to craft end-user applications. I wrote about this situation just a few months ago.
OSBI is evolving. There are suite alternatives from a variety of companies with similar capabilities and but a variety of sponsor business models. I’ll probably write about some of them — Pentaho, JasperSoft, SpagoBI, OpenI, Palo Server — soon.
May be at Gartner are too busy playing the Magic Quadrant game to notice that things are changing?
Thank you Seth, and please keep us updated!
Roberto Galoppini and 
Truster are discussing.
Truster 9:54 am on September 9, 2008 Permalink
Thanks for the post and the interesting interview you had with Seth. It might change the way I look at the Internet: I did not know it started up that early. And it reminded me the time when most people did not have access to the internet; my first browsing a few years ago.
But the purpose of this comment is not to talk about myself but to ask a question about OSBI and certain programs I have to choose. As written in the post, I have looked at applications you write about and others. So far I am pretty fond of two programs: Talend and Pentaho.
It is difficult to make a choice when it comes to the performance, the compatibility and the components of the software. But a difference can be made by comparing the GUI which might be more user-friendly with Talend Open Studio: I especially like the tmap component in Talend Open Studio, a really good way to get a graphical and functional view of integration processes. The tool also has a good debugging system and an active community able to help you. Have some of you tried it out?
Roberto Galoppini 5:44 pm on September 9, 2008 Permalink
I have just asked Seth, but unfortunately he is not a Talend Open Studio user.
Googling around there are few free resources available comparing the two applications, at the end of the day choosing the “right” application is time consuming (what I call “the cost of free”).
Keep us updated if you want, and happy hacking!