It’s hard to believe my company, OpenBI, a BI consultancy focused on implementing open source software, is approaching its fifth anniversary.
We’ve built our business by working closely with platform vendors Pentaho and Jaspersoft, and are big proponents of the R Platform for Statistical Computing, currently negotiating partner status with R commercial purveyor Revolution Analytics.
When I discuss company history with my partners, we acknowledge we might have started a couple of years in front of demand for OSBI. We did a lot of evangelizing with small companies in the early days, many larger firms hesitant about open source for a host of reasons, not the least of which were legal fears. Our early customers were, not surprisingly, small and startup companies that had been priced out of the BI market pre-OSBI.
The commercial OSBI vendors spent a good deal of time early on trying to figure out their business models. At first, the enterprise and community releases of software were essentially the same. What distinguished enterprise from community was the support the vendors provided customers for their subscription fees. Prospects were reluctant initially, so the vendors measured themselves not by traditional software sales metrics but instead by growth in the number of downloads of their free software. It was easy to tell that they were treading water by looking for what was not said in the marketing materials.
That’s changed now. Business is thriving for the main OSBI vendors, and the KPIs are revenue and profit rather than software downloads. The OSBI business model has evolved from support to software, with the for-subscription enterprise release now much more functional than its free community brethren – the so-called freemium or open core software models. Companies that seek to build serious BI apps must look long and hard at enterprise advantages.
And many large customers no longer seem reticent. Indeed, even while smaller OEM/ISV customers still represent a sizable percentage of subscription revenue, use by larger firms is growing dramatically. Two years ago, 75 percent of OpenBI’s income came from customers with $50 million or less in revenue; now, 50 percent comes from customers larger than $250 million. The benefits of adoption are being seen by the entire OSBI ecosystem. I think it’s safe to expect significant annual growth from the key OSBI vendors over the coming years.
As a sign of this maturity, both Pentaho (Pentaho Global Partner Summit) and Jaspersoft (JasperWorld 2011) are sponsoring international conferences in San Francisco this winter. I’ll post blogs on each. The remainder of this note is on PGPS, with material summarized by participating partners Dave Reinke, Bryan Senseman and Kevin Haas. Colleague Ace Lal and I will cover JW.
For Chicagoans, San Francisco is a welcome respite from January wind chill challenges. So there was no shortage of OpenBI volunteers for the Pentaho Summit at the historic Presidio Golden Gate Club, January 19-21. The conference facilities were stunning with panoramic views of the Golden Gate Bridge and the bay. Sunny skies with temperatures in the 60s made the experience especially noteworthy.
CEO Richard Daley affirmed the breakthrough year for Pentaho, reporting over 100 percent revenue growth for 2010, with more of the same indicated for 2011. He also reiterated Pentaho’s goal of commoditizing the BI marketplace with a complete BI platform. Pentaho’s product strategy will continue to highlight its data-centric orientation, leveraging the strength of its flagship Pentaho Data Integration (PDI) component as a central part of its platform.
Product Marketing VP Joe Nicholson waxed on Agile BI, the centerpiece of much company messaging in 2010. Nicholson’s point of departure is that Pervasive BI has failed, victim of complex and costly deployments and a shortage of qualified IT staff. The antidote? The reduction in time, complexity and required resource bandwidth engendered by Agile BI – and facilitated by Pentaho’s tight and seamless integration between data and reporting. Call me a skeptic, but while I think an agile approach is good for POCs, I want to see some numbers evaluating the performance of Agile BI before I anoint it the new best methodology for enduring BI deployments.
Product Manager Jake Cornelius teased participants with capabilities to come in future releases. Among the new components will be Information Quality, a tool under development by a partner that will include address cleansing, improved multi-tenancy for shared platform access, Hadoop integration, and integration of the Weka statistical learning package with PDI. OpenBI’s excited to try out the PDI-Weka demand forecasting functionality with time series methods for one of our customers as an illustration of the close ties of data integration with predictive analytics.
In his discussion of the new PDI-Hadoop integration, Chief Geek James Dixon cited transactional, sub-transactional and non-transactional uses for Hadoop as a component in an enterprise BI platform. He introduced the concept of a “Data Lake” – single source, large volume data that cannot be distilled but can, however, be treated. His take is that most big data suitable for Hadoop thrives at the sub-transactional levels of web logs, social media, sensor and financial market bid-ask. Hadoop was originally design to solve the problem of “indexing the Internet,” but Pentaho is leading Hadoop into the BI market as a processing engine for big data.
Dutch OS thought leader Jos van Dongen delivered a humorous keynote that offered positive commentary on the state of OSBI while lampooning the “Magic Quadrant”-like work of several large industry analysis firms – judging them to be in the lower left quadrant with axes of “ability to inform” and “completeness of research”! He argues that the large proprietary BI software firms now face the innovator’s dilemma with their consolidations, leaving a significant opening for OSBI vendors.
Citing Mark Madsen’s research on OSBI, van Dongen notes that lower cost, reduced dependence on vendors and ease of integration will continue to drive companies in the OSBI direction. He thinks that Pentaho, with its many successful community projects and growing ecosystem, is now a perfectly suitable answer for basic data warehousing, even as it’s still a bit short on enterprise niceties like meta data management and visualization. His recommendation to Pentaho: continue beefing up the enterprise stack while focusing at present on PDI, Analyzer and embedded applications.
Having attended each of the previous four summits, OpenBI found this one, with over 100 enthusiastic world-wide attendees, to be the biggest and best by far. The mood was upbeat throughout. Notable was the absence of cloak room commiserating among partner companies that marked early summits. OpenBI’s very bullish on the state of OSBI and Pentaho and looks forward to collaborating in 2011.