Cover Story

Data Mining & Business Intelligence >> Open for Business

• By Bennett Voyles
• 02/01/07

As open source BI attracts the attention of campus IT departments, carefully weighing the pros and cons becomes vital.

You know about the Sakai Project (open source course management system); you may even know about Kuali (open source financials). So, what’s the next wave in open source software? Today, community-source-minded developers have set their sights on business intelligence systems.

Though open source BI may still be only a rumor in most campus IT departments, some brave early adopters have begun to experiment with these applications. These pioneers say the programs often are surprisingly good, and provide much of the functionality that the big software companies promise, at licensing fees that are an order of magnitude lower, if not altogether free. Beyond the cost advantages, higher ed users of open source BI say open source programs offer them greater flexibility and control, and they are confident that these programs will keep on improving.

“My sense is that open source BI is not quite ready for prime time, but it’s getting there,” says David Wells, director of education for TDWI, a sister institution of Campus Technology. “Folks like Pentaho have done some really interesting stuff. They’re almost enterprise strength.”

Options and Freedom

At first glance, the open source route might seem like a strategy that would be of interest only to minor league schools with minimal IT needs, but such large institutions as Indiana University and the University of Nebraska are very serious about open source BI, and with good reason: Open source initiatives allow their members to “share the wealth” of collective advances in development, with a great deal less effort than it would take a lone shop to develop a new module. What’s more, as the community effort expands, applications become richer and more fine-tuned as they are tested, used, and improved by a growing community—actually, a “user group” with vested interests. Then too, such initiatives can free their members from what is often seen as the vendor tyranny of escalating costs, dropped features, even dropped products and support. And open source options are not as limited as one might think, even in this nascent period.

“There is an astoundingly rich array of choices in the open source world,” remarks Amy Stephen, a data and internet specialist for the University of Nebraska. “This is not hobbyist stuff; this is absolutely amazing stuff.”

While open source software may have an “alternative,” Birkenstockish past, most higher ed users now experimenting with open source BI aren’t coming on board simply because of the anti-bigcompany animus that had fueled some of the earlier enthusiasm for what was initially termed the open source movement. Stephen, for example, says that she and her colleagues at NU actually came quite reluctantly to the decision to use open source. “We are not religious zealots here; we are simply trying to get a job done.”

Compatibility is Key

After a dozen years as a satisfied Microsoft customer, Stephen and her teammates concluded that they could no longer use Microsoft SQL for their reporting needs. The latest version of SQL required use of Microsoft’s active directory (AD) system, but NU needed a solution based on a more open standard, compatible with the different kinds of systems the university supported. Administrators “couldn’t use the AD approach if we wanted to,” Stephen explains, adding that her team decided to use the iReports report writer and Jasper Intelligence server from commercial open source BI solution provider JasperSoft, for the same reason that they had originally decided to use Microsoft’s SQL server and NT: compatibility. she claims that JasperSoft allows users to connect to other applications regardless of their place in the server directory structure, unlike Microsoft’s high-end BI products.

Stephen says she asked the Microsoft people “a million times” why the company had created this lack of flexibility, and she came to think it was a marketing ploy. “I think they believed—and in many cases it’s true—that they could control their market by forcing people into the same active directory structure,” she says.

Though she and her team were slow to turn to an open source solution, she says it’s turned out to be “a joy and an awakening,” as she’s become aware of open source offerings not only in BI, but in all kinds of applications.

Stephen confides she has been surprised at the quality of the JasperSoft tools now being rolled out on campus, and singles out the company’s iReports module as a particularly strong one. “iReports is every bit as rich as Crystal Reports [www.businessobjects.com], and it generates beautiful reports,” she says. Shifting to JasperSoft has “made it possible for us to use high-end tools, and to replace the old, archaic, hand-carved drill-down reports that we had developed in-house,” she says.

Price and Flexibility

A lower price tag is another reason campus IT professionals are drawn to open source. Advocates of the approach insist the systems can give an IT department a way to start providing high-quality business intelligence tools quickly, without having to wait for a budget committee to write a $3 million check.

For years, cost concerns kept Indiana University administrators from purchasing a BI system. “It’s hard to ‘sell’ something people don’t know,” says Rebecca Gribble, manager of decision support services at IU. But since her department installed the open source Eclipse BIRT (Business Intelligence and Reporting Tools) system this past fall, Gribble believes she will soon build up a constituency for BI, which may make it easier to include a more established system in the budget at some point—that is, if the school even needs one; at the moment, it’s too early to tell. “We may find that it meets our BI need, period,” says Gribble. “Or, we may find that it just whets the appetite,” she adds.

But consultants and open source BI vendors cite the ability to customize the systems as another key advantage. For a campus IT shop with some in-house programming expertise, having the source code makes it possible to customize the software to meet campus business needs, rather than the reverse: revamping processes or back-burnering need in order to meet software requirements.

The University of Chicago Press (the publishing unit of the University of Chicago) is likely to become another early academic adopter of open source BI. The IT team of the UC Press is now evaluating Pentaho, a full-service BI suite and JasperSoft competitor. “We are still at the evaluation stage and it’s looking very promising,” says Matt Fisher, senior developer for business architecture. What the IT team sees in Pentaho, says Online Business Manager Andrew Brenneman, “is an ease of use and a set of objects that we could easily leverage.”

The UC Press IT team had been looking for a reporting solution for some time before it turned to open source. Says Brenneman: “Not surprisingly, we have a constant drumbeat of requests for reporting from management, and while previously that need was met with periodic reporting that included sales and transaction data, etcetera, it has become very clear to us that what senior management needs is transparency—real-time transparency—to the data that drives our business.”

In IT shops that have some development expertise, open source BI can provide some interesting new options in addition to lower price tags, points out Joseph di Paolantonio, a Montara, CA, technology consultant. He recalls the case of an IT staffer at a school whose administrators wanted a program to analyze financial aid data, and needed the software right away. Unfortunately, IT had virtually no funds to purchase the application. But the staffer discovered that his team could simply download Palo, an open source, online analytical processing (OLAP) system—essentially a multi-dimensional analytical plug-in for Excel—and have it up and running within a week.

Compare that with trying out one of the established BI names, di Paolantonio says, where you can only work with sample data. With Palo and most open source systems, “you can download any of these projects and just start using them; start your prototyping,” he maintains. If the prototype goes well, you can decide either to purchase a license (many of the private companies that act as stewards of these open source systems sell enterprise versions that include more heavily tested software and smoother functionality), or a contract for support.

Brenneman at the UC Press sees another flexibility advantage, as well. “Obviously, by not being dependent on proprietary technology, we have more latitude when it comes to potential vendors and potential service providers, which gives us greater leverage on pricing,” he says. Open source also makes it easy to work with a variety of different sources of software and data, he adds.

A free download doesn’t mean the whole system is free, however. Often, an open source provider will have a free version available for download and a more feature-rich version (including some proprietary software) available for license. Licenses for open source BI still cost money, although typically they cost only 10 to 20 percent of what a commercial BI developer would charge for proprietary software, open source providers say, and there’s no large recurring fee. (Importantly, some commercial providers do “open source” their code; the distinction here is between proprietary and open source software, whether it is commercial or not.)

On the other hand, the cost of support can still be an issue. Some CIOs reportedly have shied away from adopting an open source program, fearing that free software may turn out to be something like adopting a “free” puppy. Others with experience insist that support is generally available at costs comparable to that of support for a conventional software system. Stephen at NU says that in fact, this was a key selling point for her CIO when the school decided to begin using JasperSoft applications.

Such support can be offered either by a company that licenses the full version of the software, or by an independent consulting service. At JasperSoft, for instance, support is available for free via the vendor’s online newsgroups, or by contract. Stephen puts it this way: “You can stand in line for the city bus and then ride with the commoners, if you will, or you can get a little more handholding.”

Finally, there is what Eric Gamazon, development manager at the UC Press, calls the “intellectual” cost of using open source BI, since typically the tools take some time for the IT team to learn to use and to modify. “You have to go through the documentation and understand it, but once you have, there’s true ease of use and flexibility.”

Higher and Lower Risks

Still, if open source BI is so great, why haven’t more campus CIOs heard more about it? One reason, says di Paolantonio, is that open source software isn’t “out there,” being hawked by high-priced salespeople. Since the software can be downloaded for free or for a minimal fee, the marketing budget is much lower, he says. As a result, he adds, “it’s not real likely that you’re going to find some Armani-suited field service guy who pulls down a million dollar commission check, trying to convince you to opt for open source BI.”

Another reason: the perception by some CIOs that a commercial company is somehow more accountable, even through legal recourse (a commercial vendor, after all, can be vulnerable to litigation if its product doesn’t work). Yet, many open source proponents feel the option of legal recourse is more or less a false sense of security that doesn’t pan out in the real world anyway. (How many technology vendors has your institution sued recently?)

Open source BI is also a newcomer to the technology scene, and new software is often not quite housebroken. Consultants and vendors caution that very small IT departments may have difficulty installing the programs. “If your IT department consists of close to zero technical staff, you might have some challenges,” says Lance Walter, VP of marketing for Pentaho.

On the other hand, open source could work well for a school with a lot of student programmers, points out Mark Madsen, an IT consultant based in Rogue River, OR. “If you’ve got lots of smart young people working for peanuts, in that environment open source BI probably becomes very attractive.” Yet many open source BI programs are still fairly rough around the edges, though they are getting more user-friendly all the time, says Madsen. Open source BI “is not something that you can just pull out the box and install the CD; it’s not like that at all,” points out Rick Mortensen, CEO of Marvelit, a maker of open source BI “dashboard” software—a visual tool for campus execs who want to see key changes in business indicators at a glance.

On the bright side, in the current climate of mergers and acquisitions, open source sidesteps some of the risk that proprietary systems present. If a commercial software provider merges with another, there is nothing to stop the acquiring company from shutting down the smaller business, or encouraging migration to one product by making another product difficult to use. But with an open source system, supporters point out, most of the code is in the hands of the campus developer. In a sense, this “future proofs” the enterprise, because the community of users, not the company, has control of the software. Explains Walter of Pentaho: “You’ve almost got this automatic insurance policy: other folks like you, who are in the same boat.”

Di Paolantonio says there also tend to be fewer new releases of open source software, as opposed to commercial. After a certain point, he says, many commercial vendors continue to create new releases but the reasons have more to do with a company’s revenue stream than with improving performance for users. The open source companies, on the other hand, “are not trying to sell you the latest greatest version just to make some money and give you a new GUI,” he says.

Major Changes Ahead

Stephen believes the rise of open source is a major event, like the end of punch cards (which she witnessed while working for IBM in the 1980s), and later, the rise of Microsoft. Today, open source “is getting real,” says Stephen. “I went to the Linux conference in San Francisco in August and it reminded me of when Microsoft was first evolving,” she says, pointing out that open source is now at the stage of evolution when the Birkenstocks and ponytails disappear, and open source execs start to buy sharp suits. She believes that there are now good open source alternatives in almost every category of software. “I can’t think of an area in IT today where open source is not offering equal or better tools,” she says.

When it comes to business intelligence in particular, however, the revolution may still be a few years away. David Brower, global education market manager for Novell, believes that such an evolution may take some time: Schools that have spent millions to install commercial ERP systems aren’t going to be too eager to rip them out and start over any time soon. That will change over time as the functionality improves. All open source has to do is get close to where ERP functionality is now, and schools will begin to cross over, Brower maintains.

Right now, says Marvelit’s Mortensen, the choice between open source and proprietary BI comes down to this: Do I want a cheap solution that gets me 80 percent of what I need, but is painful, or do I want a slicker, more expensive solution, that is also painful?

Open source BI may not be perfected yet, but TDWI’s Wells advises keeping a close eye on it. Institutions that aren’t ready to take the plunge now should check back in 18 to 24 months. “But don’t wait five years,” he warns: A longterm investment in a proprietary system may make a switch more difficult.

WEBEXTRA :: CT speaks with open source visionary Lee Belarmino of San Joaquin Delta College (CA).