Enterprise Storage Forum content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
Data fabric has become one of the most commonly used, and perhaps most-hyped, technology architectures of the last few years. In fact, after Gartner named data fabric one of their top ten trends two years ago, it seems that the concept is everywhere. Yet data fabric, and the type of computational architecture it describes, is still poorly understood.
In this article, we’ll take a deep look at data fabrics, including what the term means, what these systems are supposed to achieve, and how they will affect the future of data storage and processing systems.
Also read: Storage Infrastructures for Edge Computing
What is Data Fabric?
In order to understand what data fabric is, it’s worth thinking about the way that computer systems develop over the medium term. Most engineers and networks administrators will be familiar with the idea that systems architecture changes in a cyclical manner.
Every few years, a solution will arise that claims to bring together multiple data silos into an integrated storage management platform. The ease of having all your data in one place, and one integrated tool for managing it, will drive adoption of this system. Then, due to security concerns, various parts of this system will be broken off, combined with new functionality and data stores, and siloed themselves, fragmenting the system once again. Then a solution will arise that aims to integrate everything back together, and so on.
The idea of “the” data fabric – or of “a” data fabric, at a more granular level — is to break this chain of continual fragmentation and recombination. Instead of viewing your data platform as a series of silos that must be brought together, data fabric approaches recognize the complex and important interactions between data silos, while also recognizing the value of keeping them distinct.
In order to achieve this, data fabric approaches make use of semantic graphs. These are maps of your network that can be used to generate high-level topological representations of it, but also imbued into the data it holds.
The Advantages of Data Fabrics
In some ways, the emergence of data fabrics is the consequence of an increased academic interest in network topology over the past decade. In others, it is a direct response to some of the practical problems involved in digital transformation. With so many companies undergoing digital transformation concurrently, it’s no surprise that 93% of companies have experienced problems with integrating their systems with their suppliers and customers.
Data fabric approaches promise to make digital transformation processes more efficient by applying semantic data mapping onto these systems. Organizations can fix the biggest differences at the data levels that underlie these technologies, by inserting a semantic graph integration layer over the existing structures.
This can be achieved in several ways. Organizations may choose to use data virtualization, storage tiering, or ETL, but all approaches can then be combined with semantic graph technology to track data through complex architectures, and ensure that it is available to the right services at the right time.
When taken together, these capabilities mean that data fabric architectures are likely to become the standard approach for data-intensive applications within a few short years, and that a group of semantic graph specialists are likely to see their workloads increase many times.
Also read: Trends in AI-Driven Storage
Data Storage for Data Fabrics
For most organizations, moving to a data fabric model will involve commissioning external experts to develop and then apply semantic graph data as an extra layer “above” their existing data architecture. However, there are steps that can be taken in-house to prepare for this transition, and which will make it much more efficient as it progresses.
One of the most important of these is to put in place a rigorous, secure, and adaptive system for storing data. Today, this means using a hybrid cloud model that provides easy accessibility for some data, and high levels of security for others. This is a model that has been gaining traction recently, but for many firms it may still seem that managing multiple clouds will be a complex undertaking.
There is no reason, however, why that has to be the case. Indeed, by working with a specialist in data storage, you can quickly build a network model that is both complex enough to deal with the varying demands you put on your data, and yet simple enough to manage in-house. In doing so, not only will you be preparing for a future in which data fabric will become standard, but you’ll also be building an architecture for streaming data, and therefore one that will ensure your data capabilities stay current.
The Flexibility of Data Fabrics
In short, the data fabric architecture has turned into the most economical and flexible choice for the modern data ecosystem. It’s one of the primary reasons why the top data protection software and solutions are explicitly built with data fabric in mind.
Read next: Best Storage Management Software
