|
By Lauren Bielski
In the time since last year's consolidation, the business intelligence sector is making incremental but steady gains toward the long-cherished ideal of generating "actionable insight" for a wider range of business users.
Improvements are coming from emerging niche vendors (such as open source vendor JasperSoft), as well the established top dogs and IBM, Oracle and SAP, which are incorporating the wares of BI specialists such as Cognos, Hyperion and Business Objects into their respective offerings. Certainly, everyone continues to talk about easier-to-use interfaces.
"The big consolidation wave [in 2008] has, in some sense, left a vacuum," says consultant Tom Hudock. "Emerging vendors using SaaS and in-line memory to offer capabilities could take the industry in a new, more intuitive direction." Hudock blogs about business intelligence and says the industry has been spinning some interesting changes of late.
And yet, in a recent white paper about BI trends, written by Don Tapscott, a 2007 survey of CIOs was referenced, indicating that 64% of respondents thought they didn't have access to the right information to run their businesses.
Clearly, then, progress is not perfection.
Business users are still on the hunt for novel insights about customers, operations and products. At the same time, many want tools that are easier to learn and adopt. Think of it as the quest for the "Google effect," or the ability to use simple, natural language queries to get information from applicable structured, unstructured, internal and external sources.
Instead, data at most companies mostly winds up in spreadsheets, where transformations that get done can't be fed back into transactional source systems. This is the case despite advances in databases and the existence of tools such as datamarts, which are designed to optimize storage for BI applications.
Progress, but users still want more
What to make, then, of this notion of BI-related progress that isn't moving quite fast enough? Partially, it's a matter of perspective. "Companies have been mining data for decades, but traditionally it has been a costly, and technically arduous process," notes George Tomko, a consultant to CIOs based in St. Louis who has held numerous CIO positions in his career. "These days, affordable storage options and developments such as in-line memory BI tools represent real gains in terms of what types of analysis is possible."
And yet, Tomko relates an oft-heard truism in the field: lifting the data from a given database and transporting it to the business user in a useful form is a complex, multistep process to get data in the right form. Moreover, successful users of most traditional BI systems are analysts with technical prowess, because queries need to be constructed in specific ways. And problems with shadow data and other organizational issues with respect to data management still loom large. At the same time, BI projects have tended to take about a year to deploy and have cost millions.
"The industry is setting the bar a bit high if it is envisioning ‘BI for the masses,'" says William McKnight, strategist, lead enterprise information architect and owner of McKnight Associates, Inc.
"The tools aren't yet at that level, nor do I think most organizations are set up to work that way." While McKnight thinks the findings of BI systems should be widely shared and acted on, he doesn't believe that "all hands on deck" need to have access to such tools.
But McKnight believes advances of a different type are taking place. He indicated that a cluster of vendors—including SaaS provider LucidEra; San Mateo-based Greenplum, with its self-service analytics via the cloud; and Oracle's ExaData storage server and HP's NeoView database—each offers their part in a value chain that has the potential to bring better performance with a lower total cost of ownership.
And, speaking of Google, McKnight thinks that BI vendors ought to keep an eye on the technology giant. "The company has so much data to work with and such a culture of incubating ideas and experiments to spur innovation, I would expect them to come up with something interesting."
Does the call for more analysis, and criticism of BI, mean that it hasn't worked before? Previous generations of BI tools, says Hudock, have helped with several classes of historical data analysis. Companies have received numerous benefits from improved reporting.
Yet key challenges remain. Establishing new patterns by looking deeper at interrelationships among different classes of data is still a struggle. So is doing anything predictive.
Yet Hudock is excited by some emerging vendors that are coming up with solutions aimed at internet companies. He mentioned a potential game-changer in KickFire, which supplies a mass market analytic appliance, primarily aimed at internet companies that outgrow their MySQL databases. "Delivering business intelligence with an appliance is an interesting idea."
Hudock also acknowledged the hype around business intelligence in the cloud, although he is less enthusiastic about that class of vendors at this point. "Some of it is marketing driven. We'll have to see what happens."
Closer to real-time
"We're closer than ever to real-time performance, and yet I don't know of too many organizations that run with dashboards offering today's information or this week's information," says Tomko. "Many companies are certainly looking to get that operational business intelligence."
Gartner estimated that operational BI would continue to gather momentum, and last year indicated that 90% of Global 2000 companies would have mission-critical BI and data warehouse systems, up from 25% in 2007.
"To me, what's needed are systems that can help you figure out what you don't know yet," says Tomko. "Understanding your information gaps rather than simply turning around routine queries ever more quickly should be the objective."
Data management also remains a challenge that has been often discussed. "You can have powerful front-end tools, but performance is very much influenced by the design of back-end systems," says McKnight. "Master data management is starting to take off as a discipline because firms are realizing that accurate data that is controlled is critical in making BI efforts effective."
Use of metadata—literally, data about data—as part of a master data management program can help deliver the right information to the right decision makers.
Only registered users can write comments.
Please login or register. |
