The winter gods must be smiling on me. Two weeks ago, I dodged the big Midwest blizzard by traveling to Santa Clara for Strata. Last week, I skirted the deep Chicago freeze by participating in JasperWorld in San Francisco. Back in Chicago at the tail end of sub-zero wind chills, I only wish I could have stayed a few more days.
The inaugural JasperWorld exceeded my expectations, which were admittedly modest. I generally find vendor conferences a bit too rah-rah for my taste. There was plenty of that at JW but substance as well. I left with an even more upbeat opinion on the future of open source BI and Jaspersoft. My sense is that most of the 200 participants representing 16 countries, heavy on OEMs, ISVs and consultancies, shared that view. The conference was administered flawlessly at the Hyatt, Fisherman’s Wharf.
High energy Jaspersoft CEO Brian Gentile cited a welter of “research” that corroborates an evolving network-centric, Web 2.0 BI world revolving on the requirements of Gen Y practitioners, the infrastructure wealth of the cloud, and ever-increasing cost pressures. Ubiquitous access, IT-enabled, self-service BI, and the challenges of big data are front and center to Jaspersoft’s strategy. Their annualized 50 percent growth over the past few years affirms the company’s stature in the BI world.
Keynote speaker Howard Dresner’s point of departure was a performance management culture that correlates with organizational success in the use of BI, citing his research and books. Among the BI trends he notes are an increase in cloud and SaaS deployments; the growth of mobile BI; smaller-sized company adoption of BI; the demand for smaller data latency; the enhanced use of visualization; and the evolution of collaboration frameworks to fuel the science of business.
Former MySQL rock star and current Eucalyptus CEO Marten Mickos discussed evolution of the open source LAMP development stack from just 5-10 years ago. Driving from the mantra, “The future is here, it’s just not widely distributed yet,” Mickos described a changing and challenging technology landscape that includes continued growth of the internet; the prominence of mobile computing; ever-increasing data; a higher proportion of online biz, democratic consumerization; and ephemeral, ad hoc access using ample, dynamic infrastructure resources. To support these trends the Lamp stack will evolve to one that has many more levels – but at its core is built from open source components. It will be more elaborate, more diverse and scale out significantly better than LAMP.
The Big Data and Analytics panel brought together Jaspersoft partners – database vendors MongoDB, VoltDB and Riak from Basho with statistics/data mining purveyor Revolution Analytics. RA CEO Norman Nie enthused that the large case numbers of Big Data force statisticians to expand their predictive bases beyond their limited linear models – for the betterment of BI. He cited RA’s new virtual file processing for data sets larger than memory. Each of the database platforms touted different data sweet spots, with the vendors acknowledging Big Data as defined by the recent Strata Conference is not among them. If the Big Data dbms categories are Hadoop, NoSQL and MPP, these vendors are closer to NoSQL. Jaspersoft is investing heavily in connectivity with the Big Data sources above.
Several presentations focused on Jaspersoft 4.0. The most significant change with the new version is the overhaul of the user interface infrastructure, among the resulting positives being the introduction of web application standards in the Jaspersoft BI stack. As a consequence, many customizations can now be implemented within the Jaspersoft web UI. A gallery of components and samples are available to the platform, facilitating quick prototypes and changes by user experience teams. Advanced UI changes would require file editing, but the content, presentation and interactivity have been cleanly separated into standardized JSP, CSS and JS files, ensuring compatibility with future releases. In keeping with JS’s support for multi-tenancy, the themes can now vary by tenant.
JasperWorld participants were treated to a whirlwind tour of these configuration changes with the updated interface framework. Developers can create custom parameter pages that implement advanced UI components like tabs, search-able lists and cascading parameters.
Finally, Jaspersoft 4.0 BI Enterprise Edition will ship with a comprehensive, commercial version of Talend’s Data Integration software. Having a full-featured ETL suite core to the platform is a step forward for Jaspersoft, even if it’s OEM’d from Talend. Without ETL, Jaspersoft was seen more as an embeddable reporting tool serving the OEM/ISV market than an enterprise BI platform. The Jaspersoft-Talend relationship will hopefully be seamless to JS users.
Jaspersoft founders Giulio Toffoli and Theodor Danicu entertained attendees with tips and tricks for iReport and JasperReports, respectively. Toffoli also demoed the new designer built in Eclipse while Danicu showcased interactive JasperReports to support self-service BI.
Christopher Creel from DST Healthcare showcased his company’s mobile reporting application for iPods/iPads that pulls charts from Jasperserver and displays them natively. The app allows users to share these charts via email. If the charts show a potential issue, the mobile app can open a ticket for follow-up. Users can also run and cache reports for later offline review.
Jaspersoft VP of Product Strategy Karl Van den Bergh teamed with VP Engineering Joseph Rozenfeld to lay out the Jaspersoft product roadmap. The go-forward goals are to enable better BI building, to facilitate self-service and to prepare for the new BI world that revolves on Big Data. Specific objectives are successful deployment of a Web 2.0, CSS orientation; growth of Jaspersoft in the Cloud; a 100% web-based stack; and Big Data – Hadoop, NoSQL, MPP connectivity. Tech guru Matt Dahlman deftly demoed connectivity to Big Data in the Cloud, JasperReports for the iPad and the new Eclipse-based Jaspersoft Studio.
One of the tech sessions involved integrating JasperReports with the R statistical platform. The combination of predictive modeling/data mining and BI is near and dear to me, a stats guy by schooling. The presenter showed in painstaking detail using PL/R – R embedded within Postgres – how to invoke R time series forecasting code within a JR session, save the output, and use the generated predictions for subsequent JR dashboards and reports. Pretty cool stuff – but complicated.
Simpler is the server integration demoed by Revolution Analytics, connecting their commercial dialect of R to JasperReports. Very nice interface. The separation between the R and Jasper sides is clean. The problem for both companies right now is that it runs on the RA’s commercial R only, currently deployed by less than 1 percent of R users. I suggested to RA that at this early stage it might want to promote the product to non-commercial R users as well to help generate market share for their enterprise products. I’m surely pulling for its success.
No more conferences. I’m now stuck in Chicago for 3 weeks.