Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

The top 6 use cases for a data fabric architecture

Enterprise data fabric adoption has been on the rise as a way to ensure access and data sharing in a distributed environment. Here are the top use cases for data fabrics.

Enterprises are turning to data fabric architectures to provide a holistic view of their data. There are different visions of how exactly to go about this, but at its core, everyone seems to agree it goes beyond data lakes, data catalogs and data virtualization to provide a more coherent integration tier.

Gartner cited data fabric architectures as one of the top 10 trends in 2019 because it enables seamless access and data sharing in a distributed data environment.

"A data fabric is an integrative approach to collecting and connecting enterprise data that stresses singularity in managing, distributing and securing data," said Brian Platz, co-CEO and co-founder of Fluree, a blockchain data management platform.

A data fabric architecture provides a step forward in maximizing the value of information spread across data silos. It also provides a goal that the organization can align to provide streamlined and secure access to data in an otherwise complex distributed network environment.

Data fabric benefits

A data fabric architecture promises a way to deal with many of the security and governance issues being raised by new privacy regulations and the rise in security breach incidents.

"By far the largest positive impact of a data fabric for organizations is the focus on enterprise-wide data security and governance as part of the deployment, establishing it as a fundamental, ongoing process," said Wim Stoop, director of product marketing at Cloudera.

Data governance is often seen in isolation, tied to a use case like tackling regulatory compliance needs or departmental requirements in isolation. With a data fabric, organizations are required to take a step back and consider data management holistically.

This delivers the self-service access to data and analytics businesses demand to experiment and quickly drive value from data. Such a degree of management, governance and security of data then also makes proving compliance -- both industry and regulatory -- more or less a side effect of having implemented the fabric itself. Although this is not a full solution, it greatly reduces the effort associated with adhering to compliance requirements.

Data fabric challenges

Platz cautioned that there is a wide gulf between a vision for a perfect data fabric and what is practical today.

"In practice, many first versions of data fabric architectures look more like just another data lake," Platz said.

Folks that are building a data fabric for the first time do not account for the need for inherent data interoperability. Disparate systems will format data differently. Data that does not adhere to a global enterprise schema will not inherently speak the same language. This lack of native interoperability will add friction to the time-to-value for data stakeholders and introduce the need for harmonizing, deduplicating and cleansing data.

An organization needs to be able to understand its data consumption and regulatory and compliance needs in order to make proper use of its data fabric.

"Not understanding one or all of those areas often creates challenges or points of failure," said Morten Bagai, CTO at Grax, a data backup, archive and recovery service.

Once organizations sort out these issues, they can begin exploring new data fabric use cases such as the following.

1. AI data collaboration

A data fabric architecture can provide AI engineers with access to broad, integrative data for better-informed decisions.

"Because AI needs broad access to high-integrity data, a data fabric can support the efficient delivery of information to AI applications for quick, well-informed decisions," Platz said.

He also said the architecture is being used to enhance the delivery of AI applications for detecting fraud and building faster models for predictive analytics. For example, predictive maintenance requires streamlined access to real-time data that a data fabric provides.

2. Enhancing security

A data fabric can also improve security applications by tying together data and applications from across physical and IT systems, said Rajiv Kanaujia, vice president of operations at CloudCheckr, a cloud management platform.

For example, a team could improve security by tying together information from key readers used to open doors, which could be correlated with event data from computer systems accessed from within the facility. This would make it possible to perform more sophisticated analysis of typical and anomalous behavior to trigger real-time security alerts when required.

3. Creating a holistic customer view

Organizations can also use a data fabric architecture to weave together data from a customer's activities along with the various roles that interact with them to get a more holistic view, Kanaujia said. This could incorporate real-time data of various sales activities, potential revenue realization, customer onboarding time and customer satisfaction metrics.

For example, this might start with CloudTrail logs captured about a customer's use of SaaS services on Amazon, incorporate data from customer support requests and coordinate with new sales activities. The data fabric makes it possible to correlate across these different data sources to drive better analytics and offer useful recommendations.

4. Improving business understanding

Enterprises can also use data fabric to create a more holistic view of the business across activities and departments.

"A data fabric is critical to understanding any and all changes happening in a business over time," Bagai said.

He said it is useful to think of the data fabric as a topological map of anomalies, inflection points and business outcomes across the enterprise. This makes it the perfect training and testing set for machine learning and AI use to understand the business. This can also make it easier to implement process mining projects that make sense of business processes spanning multiple applications.

"Our biggest disillusionments with AI and [machine learning] actually originate from the fact that we often do not feed the full data fabric into our training sets," Bagai said.

5. Simplifying predictions and triggered actions

Data fabrics can also be used to train, configure and deploy simple prediction algorithms and trigger actions that run across various enterprise application endpoints. These types of use cases span everything from security traceability to audit compliance and revenue-generating events like cart abandonment action, ad optimization, customer retention, marketing and even orchestrated selling.

"Data fabrics are going to transform the fundamental way in which businesses learn from their past and evolve over time," Bagai said.

6. Creating a data marketplace

Enterprises implementing a data fabric architecture can also set up a more accessible data marketplace that makes it easier for citizen developers to weave disparate data sources into new models. A data marketplace allows data engineers to set up an infrastructure that can be used across multiple use cases rather than creating fresh infrastructure for each use case individually.

"Rather than implementing business use case-specific data stores or lakes to address challenges like customer churn, predictive maintenance or fraud prevention, a data marketplace addresses the data needs for the enterprise as a whole and tackles both current as well as future data requirements," Stoop said.

Next Steps

5 principles of a well-designed data architecture

Dig Deeper on Enterprise data architecture best practices

SearchBusinessAnalytics
SearchAWS
SearchContentManagement
SearchOracle
SearchSAP
SearchSQLServer
Close