The Open Source Business Intelligence Tools Are Coming!
BIRT, Mondrian, and Pentaho headline the list of scrappy open source BI newcomers.
More than two decades after the launch of the GNU’s Not Unix (GNU) software project—and almost a decade after the term “open source” was first used to describe GNU and other open software licenses—open-source software is an established model that’s every bit as legitimate as its closed-source counterparts.
Open source uptake has occurred more rapidly in some markets than in others, however. Consider the business intelligence (BI) space, for example: to date, there isn’t an open source BI analogue to Apache (or Tomcat, or JBoss), OpenOffice, Python, or other prominent open source software projects. What’s more, BI vendors say there’s been little demand for versions of their software designed to run on Linux or other open source operating systems.
That could change, however—perhaps starting this year.
Most folks are familiar with the Business Intelligence Reporting Tool (BIRT) for Eclipse, which—admittedly—is open source BI’s closest analogue to Apache, at least in terms of name recognition or star power.
BIRT 1.0 effectively went gold last June, and BIRT main mover (and prime sponsor) Actuate Corp. touted a version 2.0 release of BIRT just last month. At the time, Actuate and the Eclipse Foundation also trumpeted a BIRT partner ecosystem that includes the landed aristocracy of the software world (e.g., IBM Corp.) as well as scrappy newcomers—many of them open source BI pure plays.
The truth, says Mark Madsen, a consultant, data warehouse manager, and member of The Data Warehousing Institute’s (TDWI) research collaborative, is that there are actually quite a few open source BI projects of note.
Take Mondrian, a Java OLAP engine that supports OLAP standards such as multidimensional expressions (MDX), XML for Analysis (XML/A), and Java OLAP Interface (JOLAP). Of course, the OLAP space isn’t exactly bereft of competitors. There’s Microsoft Corp. and Hyperion Solutions Corp., for starters, along with Oracle Corp., Applix Corp., and a host of other OLAP players. And both Microsoft and Oracle bundle OLAP capabilities with their flagship—and all-but-ubiquitous—relational databases. Why would organizations embrace open source OLAP when they could just as easily tap low-cost solutions from Microsoft or Oracle?
For one thing, they might be using data sources other than SQL Server or Oracle, Madsen deadpans. There’s an even more important consideration, however: Mondrian—like the BI Reporting Tool (BIRT) for Eclipse—is becoming the BI tool of choice for J2EE software development. “[BIRT and Mondrian] are being used by all of the IT staff or people building projects using Java, so the enterprise Java people really like these things.”
Mondrian is just the tip of the proverbial iceberg, however. There’s also a full-fledged open source BI stack, Pentaho, that proposes to consume code from other open source projects—such as Mondrian. Pentaho is currently in its second round of venture capital seed funding and has hired veterans from several established BI players to manage its operations. If or when it becomes a reality, Pentaho could give J2EE codejockeys—or independent software vendors (ISV), for that matter—a combined OLAP and reporting stack, complete with dashboards and integrated workflow management capabilities.
It’s still early, but Madsen thinks there’s a good chance open source BI tools could develop into viable alternatives to low-end vendors—and to emerging commodity BI powerhouse Microsoft, too.
“It will be awhile before BI alternatives are ready, other than as embedded elements in applications. They're pretty basic, but I could see them harming the low end vendors, and offering an alternative to the Microsoft way,” Madsen indicates. “Microsoft's platform is getting too heavy and has too many forced dependencies. Still, they're the new IBM so they won't be going anywhere for quite a long while. I haven't heard anyone say the old ‘nobody got fired for’ saying about them yet, but I know that's what a lot of IT managers are thinking.”
Whither Linux-based BI?
Open source BI solutions are still gestating. Linux-based BI solutions, on the other hand, have been available for some time now: most major BI vendors offer versions of some or all of their products for use on Red Hat, SuSE, and other Linux distributions. And some BI vendors—such as Cognos Inc.—support versions of Linux running on non-Intel chips, such as IBM’s Power processors. The products, then, are there—but Linux-based BI seems to be a tougher sell.
Consider the case of MicroStrategy Inc., which offers a 64-bit version of its MicroStrategy 8 BI suite that’s been recompiled to run on Intel’s low-cost 64-bit Xeon and Advanced Micro Devices Inc.’s low-cost Opteron chips. (A number of other BI vendors market 64-bit versions of their software designed for Intel’s comparatively more expensive—and much more complex—Itanium 2 processor. The two architectures are not reconcilable.)
MicroStrategy hasn’t yet shipped a similar version of its BI suite for Windows Server 2003 x64—Microsoft’s 64-bit Xeon- and Opteron-capable operating environment. That could and probably should translate into greater-than-average demand for Linux-based versions of MicroStrategy’s BI tools—at least among customers that have outgrown the capabilities of their existing 32-bit stacks. (Given MicroStrategy’s ROLAP-based toolset, its customers—more than those of other vendors—are likely to have done just that).
That hasn’t been the case, however, says Sanju Bansal, MicroStrategy’s chief operating officer. “As soon as organizations are able to move to 64-bit, they’re very eager to do so. We saw a lot of movement in 2005 to 64-bit Solaris or AIX,” he comments. Ironically, says Bansal, 64-bit adopters seem to prefer AIX or Solaris over Linux—which not only runs on cheaper hardware, but is typically cheaper to support, to boot. “I don’t have any data, but my intuition from what I’ve seen and my talking with my technology people is that right now there is more interest for Unix adoption in the Solaris and AIX world than there is in the Linux world,” he comments. “The folks who are looking for the low cost BI solutions, they’re most often deciding between Windows and Linux anyway, and my sense is that right now, they’re reasonably happy with Windows, even though it’s not 64-bit—they’re happy with 32-bit.”
Many BI players report similar experiences: there’s demand for Linux-based BI solutions, to be sure, but it’s typically only a fraction of that for Windows-based BI—and, moreover, is often significantly less than demand for Unix-based BI solutions. Madsen, for his part, says there’s probably something to this. Windows, right now, anyway, remains the low cost BI leader—such that organizations don’t mind paying a premium (relative to Linux) for even a 32-bit Windows stack. If anything, Madsen suggests, most of the momentum will be on the open source database front, instead of Linux.
“[T]here's not as much demand for ports to Linux. I suspect the real demand on vendors is for open database support more than worries about platforms. The cost differential is significant enough, and the products are equivalent enough, that it's really crossing over,” he comments.
Indeed, open source databases such as MySQL or PostGREs need no introduction. MySQL, in particular, is doubtless familiar to a lot of BI pros through its relationship with Business Objects SA, which bundles a free version of the open source database for use as a repository with Linux- and Unix-based versions of its BI and reporting platform. Madsen, for example, says his current client is successfully exploiting the MySQL and Business Objects combination. (http://www.tdwi.org/Publications/display.aspx?Id=7809)
Once again, it’s still early, but Madsen thinks open source databases such as MySQL could reap a windfall of sorts. “I was doing the math on the license [and] support, and even with an added FTE [MySQL] still works out versus Oracle and DB2 on an eight-way server,” he explains. “The big thing I see is the removal of other costs and burdens like license compliance and asset management, freedom to move to the OS and hardware of your choice with no restrictions, no scale out [or] scale up costs, and no worries about vendor-forced upgrades. I've faced every one of those at one time or another.”
Stephen Swoyer is a Nashville, TN-based freelance journalist who writes about technology.