I recall a senior executive of one of the world’s largest consumers of data proudly proclaiming in a much-celebrated announcement that went something like this: “We’ve made the bold decision to migrate our data to the cloud. This move will dramatically improve our information sharing because the contributing data will be stored in a common environment.” That was in about 2012, a long time ago in IT years! No one seemed to notice, however, that it’s not the physical location of data that enables “information sharing.” What happened is that multiple “data clouds” stood up, and data within each of these clouds was still isolated. That is, myriad data collections resided in the cloud(s), but were not related; they remained disconnected. One could say “data islands in the ocean called the cloud.”
Fast forward, we’ve seemingly realized that we did not achieve the kind of information sharing we (still) seek. What we seek is simple and secure access to any and all available information. The more connected our data is, the more informed the answers to our queries, the more accurate the answers, and the more effective our decision making. Thus, the advent of “information fabrics.”
The word “fabric” suggests a cooperating, interconnected and organized set of elements that features a patterned and predictable material. In an information fabric, the elements are the concepts represented in the data and their salient relationships.
In the current environment, users and automated clients pose questions to a fractal set of tables in the data cloud(s), not a fabric. When a question is asked, “line-of-business” applications query a set of pre-wired “schema-bound” data sources. The full set of available data sources and other content repositories are not queried. Therefore, although the answer to a basic question may reside “in the enterprise,” the client will never know. That is, we don’t know what we know. To be fair, even a coherent well-connected information fabric may not know all there is to know about a given question; however, such an environment would rapidly and completely provide “all that we do know”. In other words, we could indeed say, “We know what we know.”
So how do we arrive at such a state? How do we create such an environment? The answers lie in a variety of areas, including people, technology, policy, and operations. From the technology perspective, creating a semantic layer – which features schema-less, self-describing, and link-able data models, methodologies and standards – liberates content from brittle, application-specific schemas, and yields agile information fabrics. The semantic layer provides the elements that interoperate, or link, data from diverse sources to create a data and information ecosystem. Approaches that lack the semantic layer are insufficient for the task.
Semantic technologies and metadata management explicitly enable information fabrics wherein data can be used and shared across applications and domains, seamlessly. Semantic layers present enormous opportunities. Increasingly autonomous software will capitalize on “fabrics of meaning” to create intelligent enterprises, which free humans to focus increasingly on higher order reasoning tasks; that is, the mission.
Semantic layers increase data interoperability, and provide organizations greater “data agility.” For example, changes to user interfaces do not “break” the information fabric; the interfaces become model-driven. Data is portable across otherwise disparate systems, services and applications. Humans and software processes can unambiguously correlate and interpret information across multiple sources. There is no requirement to “pre-model” data for queries, because the entire information fabric is available to the query agent. Selected data do not require “syntactic matches” for the software to discover and return the answer. Information fabrics built using semantic layers transcend syntax and structure and operate at the concept level. The result: global access to information on demand, which supports decision-making.
But still, this is not a universal remedy. Creating and institutionalizing common vocabularies or models is daunting, to say the least. However, current deployments of semantics-enabled information fabrics serve as demonstrable, large-scale examples of how semantics connect numerous and varied data sources.
In conclusion, creating a semantic layer of related information is crucial to realizing the ability to “know what we know,” and to discover things we didn’t know. Creating an information fabric that connects and relates data allows applications to enjoy a global view of information. Enterprise information fabrics will serve as a pillar to create coherent systems that will rapidly and completely provide all available, relevant information on demand. It’s difficult to understand how industry can latch onto the word “fabric” if the elements (data) are not interconnected; and I’m not sure what else we’ll use to create the interconnections, if not semantics.
To learn more about how semantic technology allows companies to draw conclusions and predictions from their data faster, download the Bloor Group report “The Data Fabric Journey: Cutting to the Chase with Anzo“.