BI/Analytics Forecast Needs a Closer Look
Some claim business intelligence has had its day. Not according to Gartner.
- By Steve Swoyer
- March 16, 2016
According to market-watcher Gartner Inc., there's still plenty of gas left in the car. The business intelligence (BI) car, that is.
BI's staying power is a tendentious topic. With so much hype about big data, machine learning, and advanced analytics, some claim that BI has had its day.
Not according to Gartner, which projects that the worldwide market for BI and analytics will grow at a 5.2 percent clip in 2016, reaching almost $17 billion.
On the other hand, Gartner seems to have drastically enlarged its definitions of "BI" and "analytics."
To wit: its "modern business intelligence analytics platform" -- or BI&A, for short -- incorporates features and capabilities associated with old and new technologies alike. For example, in the older, data warehouse-driven BI paradigm, data sources first had to be profiled and modeled before they could be persisted into a repository. In new the BI&A paradigm, Gartner notes, schema can be derived on access, which means data can be persisted in flat files or flat tables. In most cases, this eliminates the need for upfront profiling and modeling.
Similarly, Gartner believes the legacy (or "IT-produced") data integration paradigm will give way to what it calls an "IT-enabled" paradigm. In this model, data scientists, analysts, and other savvy consumers use self-service data prep tools to prepare their own data sets and provision their own data sources. In these and other cases, IT becomes more of an enabler than a producer or provider. Meanwhile, the work of preparing, provisioning, and "doing" BI and analytics shifts increasingly to the line of business. "The shift to the modern BI and analytics platform has now reached a tipping point," said Ian Bertram, managing vice president at Gartner, in a prepared release.
"Organizations must transition to easy-to-use, fast and agile modern BI platforms to create business value from deeper insights into diverse data sources."
The rub is that Bertram and Gartner don't say precisely what is meant by an "easy-to-use, fast and agile" BI platform, however. As one of its rivals (Forrester Research) notes, the BI platforms that tend to be most associated with ease of use -- e.g., Qlik Sense, Spotfire, Tableau, and others -- also tend to be impoverished with respect to critical data management features.
In painful point of fact, ease of use, speed, and agility mean very little absent context.
The same BI product that's easy to use, fast, and agile in one context will become much less easy to use, much less speedy, and much less agile in another. Imagine a greenfield BI implementation, with no prior BI systems and a limited number of data sources. Conversely, consider an existing environment with an application and/or database monoculture -- e.g., an all-Microsoft or all-Oracle environment. Both settings might be consistent with ease of use, speed, and agility. Few if any Global 2000 organizations fit this bill, however. Most medium-sized or larger enterprises don't, either.
In any case, how we purchase, deploy, manage, and use BI is changing. Think of Gartner's "easy-to-use, fast and agile" formula as an aspirational metric: the industry probably won't ever get there, chiefly because the problems that need to be solved are over-determined, interconnected, and in critical ways intractable.
However, BI and analytic technologies are becoming easier to use and less dependent on resource-consuming (and latency-inducing) IT interventions.
The shift to in-memory and NoSQL is transforming the traditional data warehouse-driven BI model, which mandates that data must be modeled (and the kinds of questions business people want to ask identified) upfront. Now several traditional (i.e., relational) database platforms support NoSQL-like features such as late-binding (models can be imposed or derived at query-time) and object storage, such as the ability to (1) persist JSON objects as objects in relational columns and (2) to query against them.
A more popular approach is to shred a JSON object into rows and columns. This makes sense from a data management perspective, but is less than ideal from the perspective of application developers, who must use a technique called object relational mapping(or ORM) to reconstitute said objects. If nothing else, shredding from JSON to relational complicates data exchange in the RESTful cloud. In the same way, self-service products, from BI front-end tools to data prep tools, really are easier to use, as well as more efficacious in use.
To the extent that these tools are complemented with a well-managed BI infrastructure -- one that promotes both the laissez-faire self-service business discovery use case and the no-less-critical operational and production reporting use cases -- BI and analytics become easier to use and more agile while also being (sensibly) governed and manageable.
Stephen Swoyer is a technology writer with 20 years of experience. His writing has focused on business intelligence, data warehousing, and analytics for almost 15 years. Swoyer has an abiding interest in tech, but he’s particularly intrigued by the thorny people and process problems technology vendors never, ever want to talk about. You can contact him at firstname.lastname@example.org.