Data-Driven Enterprises: Redefining The Problem Statement
Lalit Ahuja is the SVP of Buyer Success at GridGain Programs.
getty
Enterprise decision-makers worldwide agree there’s large potential worth of their knowledge that’s nonetheless ready to be harnessed. With the huge growth in knowledge over the previous couple of a long time, many makes an attempt have been made to construct the instruments and processes essential to unlock the total worth of enterprise knowledge. But even right now, many enterprises are nonetheless struggling to be really data-driven—that’s, in a position to course of all related knowledge appropriately and in a well timed trend to make higher enterprise choices.
The Previous Strategy No Longer Works
Maybe the explanation these companies are nonetheless struggling is that the business continues to border the issue the improper approach. Let’s take a look at a number of of the technical and purposeful challenges and hurdles that enterprises say are blocking their means to turn out to be data-driven.
Technical
• Knowledge Silos: Knowledge sits in several silos with totally different codecs, constructions (or lack of construction), safety protocols and interplay mechanisms.
• Monoliths: Monoliths nonetheless exist, and accessing knowledge inside them requires extra than simply understanding the best format, API or knowledge construction.
• Actual Time: Enterprise wants and use circumstances have developed, and real-time entry to the most recent knowledge is crucial for assembly these new necessities.
Non-Technical
• Knowledge Governance: Knowledge governance challenges embody course of limitations, resembling the lack to successfully transfer, course of, manipulate and embellish knowledge to satisfy new use case necessities. There are additionally knowledge lineage and high quality points, particularly round who’s chargeable for sustaining knowledge high quality and integrity.
• Knowledge Stewardship: Companies proceed to wrestle with defining and imposing knowledge entry insurance policies. Who owns the info? Who has modifying rights? Who has viewing rights underneath what particular circumstances? What occurs when workers change positions or depart the corporate?
These challenges usually are not by any means new, and expertise options supposed to deal with them exist: knowledge marts, operational knowledge shops, knowledge warehouses and grasp knowledge administration (MDM), in addition to all types of connectors, adapters and plugins. Extra just lately, methodologies like CI/CD and DataOps are being utilized to knowledge to assist deal with a few of these technical and non-technical points.
But most enterprises are nonetheless not data-driven. Why? As a result of the issue assertion these options have been making an attempt to deal with is put all knowledge in a single place—in some kind of canonical construction with controls round it.
Nevertheless, this method has induced extra issues than it has mounted. The necessity to standardize, cleanse and consolidate knowledge right into a single big knowledge retailer is complicated and time-consuming, which is ill-suited to real-time necessities—to not point out that it creates a administration nightmare. Modern knowledge lakes are a valiant try at an answer, supporting the administration of loosely coupled or decoupled knowledge for advert hoc evaluation, however they nonetheless don’t help right now’s real-time enterprise use circumstances that depend on always up to date knowledge, together with on-premises, in a number of clouds, on the edge and streaming in from a number of sources, together with knowledge providers and companions.
The true problem is making knowledge accessible to consuming functions wherever the info lives—at extraordinarily excessive speeds and for enormous quantities of information—so invaluable insights might be extracted shortly from probably the most up-to-date model, ideally the one model. So as a substitute of making an attempt to centralize knowledge and govern it there, the thought is to decentralize and federate it.
Reframing The Downside
With this background, let’s reframe the issue assertion: “Enterprises wishing to turn out to be really data-driven want real-time entry to related knowledge throughout the enterprise and from exterior sources in a approach that enables them to construct cross-sectional views and analyze the info to make knowledgeable choices.”
Be aware, this method doesn’t suggest, require or implement any specific construction, format, protocol or entry control-related constraints. The one constraints are the necessity to entry knowledge from a number of inner and exterior sources in actual time, and the power to curate and analyze related sections of this knowledge.
This, too, will not be a novel idea. Applied sciences, methodologies and architectural patterns that allow it in some type or one other have emerged up to now few years. In truth, enterprises which have been in a position to undertake such patterns and applied sciences are nicely on their option to changing into data-driven. But we nonetheless have some methods to go.
In subsequent articles on this sequence, I’ll dive deeper into the underlying challenges related to this new drawback assertion and discover some approaches to addressing them.
Forbes Know-how Council is an invitation-only neighborhood for world-class CIOs, CTOs and expertise executives. Do I qualify?
