Lest there be any confusion, platform-as-a-service (PaaS) player Birst Inc. says that it is the biggest name in Network BI. Officials used the company’s recent Analyst Day Virtual Event to make the case for Birst’s leadership in Networked BI — as well as to educate analysts about what Networked BI is.

analysis_birst1You couldn’t blame analysts for being confused. Networked business intelligence (BI) isn’t a recognized category. There isn’t any such thing as a Networked BI Competency Center, for example. There might not ever be. Birst’s vision of Networked BI gets at a real and pressing problem, however.

It begins with the reality of information siloing, albeit of an unusual kind. We’re used to thinking of a silo as an un-integrated or unmanaged data source: a proprietary system, a spreadmart, etc. Traditionally, only sources were siloed. Now practices are, too. Birst officials cite the collision of top-down and bottom-up BI practices as the catalyst for a new, as-yet-unaddressed phenomenon.

There’s traditional, IT-provisioned BI, on the one hand, with its emphasis on process, repeatability, governance, and — above all — centralized control. On the other hand, there’s a self-service insurgency — represented by products such as Tableau and Spotfire — that gives less priority to process, repeatability, governance, and centralized control.

These two silos aren’t going away, Birst officials say, nor should they.

Networked BI, as envisioned by Birst, is a technology vision for accommodating both centralized and self-service BI. When Birst CEO Jay Larson describes it, Networked BI sounds somewhat like data virtualization (DV).

DV works by creating the equivalent of a data fabric — an abstraction layer that’s laid over data sources that are otherwise logically or physically separate — which can be used to provide a single logical view of all data. Larson told analysts, “Networked BI gives you the ability to connect all of your users and all of your data across the enterprise through an interwoven fabric of data.”

This interwoven fabric, Larson promised, will permit Birst customers to share and provision information without the “need to create separate instances … or the need to copy and move data.”

Larson wasn’t haphazardly using a metaphor. Chairman and cofounder Brad Peters also used the term “fabric” to flesh out what Birst means by Networked BI. The idea, he says, is “to create a fabric of data across an organization, some of which may be centrally managed, some of which may be managed in a decentralized way … some of which [may be] managed by divisions.

“All [silos] are connected by an enterprise fabric of metadata. At a high level, fundamentally … [this] requires … an approach that allows both for enterprise data infrastructure and user-oriented data.”

This fabric is enabled by means of an overarching semantic layer, Peters said.

Birst’s semantic layer permits users to connect to certain types or categories of data in situ, i.e., on the systems where it lives. Alternatively, they can use more traditional data integration tools and techniques to prepare other types of data. Once they’re prepared, these sources can also be exposed via DV. “For data [that] exists in place that is analytically ready, let’s put a semantic layer on top of it … so we can query that data in place,” Peters said.

“[For] other data sources for which additional prep is needed … let’s ingest that [into a warehouse-like repository] and do the prep and tie both together with a semantic layer,” he explained.

Again, this sounds a lot like data virtualization. Peters sought to downplay the similarity, however.

“Data virtualization focuses on generic virtualization [of] transactional data,” he claimed. “All of this sits in a multi-tenant environment. In a single customer of Birst, different organizations or individuals have the ability to have their own sandbox, their own place to play in.”

Peters says that the networked BI model gives adopters “this ability to blend both a user-oriented approach that’s based around a specific use case … [with] an aggregation path or analytical view that’s around specific use cases.” With networked BI, he argued, Birst recognizes that “not … everything has to be loaded into a single spot, but … that data lives in a variety of [contexts].”

There are two things to examine here. First, Peters’ critique of DV isn’t remotely true. Yes, an early use case for DV’s predecessor, data federation, was as an enabling technology for frequently refreshed reporting. (Because federation technology could connect directly to operational systems, it permitted reporting against live operational data.) DV, on the other hand, is federation reborn — and reimagined.

Today, DV is used to knit together operational systems, relational databases, data warehouses, and, increasingly, REST-ful cloud services. DV not only gets at strictly structured data sources, but at polystructured sources such as NoSQL repositories and REST-ful applications, too. Some DV technologies — such as those marketed by Cisco Systems Inc. and Denodo Inc. — consolidate data quality (i.e., profiling, cleansing, matching, de-duping), master data management, and a number of other critical quality or governance features.

In other words, it sure sounds as though Birst implements the equivalent of a DV-like synthetic architecture. This isn’t in any sense a criticism of their model.

The second thing to consider is Birst’s reinvention of the term “multi-tenant.” Not only is it consistent with the company’s PaaS BI model — multi-tenancy is, after all, the backbone of the cloud — it’s also a thought-provoking reimagining of BI in the age of self-service.

In the cloud, multi-tenancy describes a scheme in which multiple subscribers simultaneously share access to the same virtualized resources.

Birst’s pitch with networked BI gets at something similar. It permits different kinds of consumption and usage paradigms — BI reports, alerts, and dashboards on the one hand, spreadsheet and self-service data discovery tools on the other — to share access to the same virtualized data resources. In networked BI’s reconceptualization of multi-tenancy, data and analytics are virtualized along with hardware resources.

The upshot is that information consumers — be they people working singly or in business domains — are to some degree insulated from one another. A marketing analyst can mix the cleansed, consistent raw data that’s exposed by Birst’s DV-like layer with data from external and/or hosted sources. What she does with this data won’t affect other consumers.

The raw data she’s using will refresh over time, too, to reflect upstream changes.

“You’re now seeing a business intelligence application mirroring very closely what happens in an organization today. People are empowered to work on their own by leveraging and extending the work of others. [They have] new capabilities, [such as] the ability to ingest and prepare data on their own … [they’re] bringing in external hosted data, [they have] the ability to certify measures … and understand where they’re coming from … you can get to a model like this,” said Pedro Arellano, senior director of product strategy with Birst, during his presentation.

“This, in my opinion, completely destroys this notion that agility and governance are at odds with one another.”

Part of this vision from Birst is forward-looking. During the NDA portion of its analyst briefing, for example, Birst demoed several features from its upcoming Birst BI refresh. The data prep scenario described above will require Birst to expose the equivalent of a managed data prep facility to self-serving analysts.

The data an analyst uses in preparing her data set must be refreshed at regular intervals to reflect upstream changes. (At the very least, the portion of data that’s provided by Birst’s DV-like layer must be periodically refreshed.) These features are coming, Birst officials promised.

The name “networked BI” itself is far from imaginative or inspiring. The same could be said about terms such as “big data” — that most vacuous of all terms — or, for that matter, “business intelligence” itself. Unimaginative terms are the rule rather than the exception. The vision that networked BI describes, on the other hand, is intriguing. It’s a synthetic architecture that addresses traditional (centralized), non-traditional (self-service), and emerging (big data, IoT) use cases. Bully for Birst.

The preceding article was originally published on May 13, 2016 in Upside, the TDWI blog.