“If you do nothing else, make your data FAIR,” Stoop said. The four pillars behind the acronym emphasize practices that make data more easily shared between devices when needed, and more readily accessible and intelligible to people across the organization.
Click to explore
WHAT IS
FAIR DATA?
FINDABLE
“Aim for self-service that lets end users find what they are looking for,” said Stoop. Findable data can be persistently, consistently and uniquely located, and its metadata should be detailed enough to ensure that no two objects can be mistaken for one another.
F
FINDABLE
A
Accessible
I
Interoperable
R
REUSABLE
HOW DO YOU MAKE DATA FAIR?
To succeed, the FAIR methodology must be backed by the right processes and infrastructure, deployed wherever an organization’s data resides. For data to be valuable, “you have to make sure it’s FAIR everywhere: at the edge, in the data center, in the public cloud and across hybrid clouds,” Stoop said.
That’s why a is key. Rather than requiring data to live in a centralized location, a data fabric “lets you dynamically orchestrate disparate data sources,” allowing users to perform analytics on any data set, in any location, at any time, said Stoop. “[It] delivers data to where it is needed, regardless of where it is born.”
A world-class data fabric does four things to make this possible:
BUILD A DATA CATALOG
A data catalog is essential to making data findable, going beyond searchability to add meaning, “so you know what your data represents,” said Stoop. A catalog provides a starting point to access all organizational data, both human-created and automatically discovered assets. Using glossaries, catalogs provide context by adding definitions and quality descriptions to raw objects. They also promote trust in data by tracking its lineage.
IMPLEMENT SECURITY, GOVERNANCE & LINEAGE
A valuable data regime is one which enforces proper access controls in a consistent, repeatable and explainable fashion. Establish how encryption and controls are applied while data is in motion as well as while it is at rest, both on standalone devices and in different elements of a hybrid cloud. Metadata should include governance rules that dictate how data must be transmitted and stored in a manner compliant with relevant rules and regulations.
ENSURE OBSERVABILITY
Proper storage and access controls are only the beginning of a healthy and FAIR data fabric. “You need to know and understand how your data is used,” Stoop said. This includes tracking the health, utilization and support status of data platforms, the cost and performance characteristics of user access to data and the behavior of user populations interacting with data. These observations promote understanding of data’s value and make it possible to take action faster when user experience or data infrastructure strays from the norm.
PROMOTE REPLICATION
Today, data is rarely used only in the location or on the device where it is created. Proper, comprehensive data replication spans traditional disciplines like backup and disaster recovery management as well as complex multi-tier application development environments and rapid cloud deployment. Data cannot simply be copied; it must be correctly and completely replicated along with all relevant metadata.
HERE’S HOW 3 COMPANIES DID IT
FAIR data principles and practices pay off for enterprises in multiple ways, from internal operational performance improvement and cost reductions to innovations serving end customers.
Below, click to hear Stoop explain how Cloudera customers across three industries used the company’s hybrid data platform to establish data fabrics that deliver consistently FAIR data to its users, expanding access to data-driven insights and actions. Each one achieved returns with a tailored architecture, but all leveraged Cloudera Data Platform capabilities to get more out of their data.
Listen To The 3 Case Studies
BANKING
CREDIT
Healthcare
How Regions Bank’s data fabric reduced fraud alerts by 30% and delivered $10 million in annual customer retention benefits
How overcoming data sprawl helped a consumer credit organization expand into new markets, while complying with global security regulations
How healthcare analytics provider IQVIA built a fabric so effective at navigating complex healthcare data operations, they now offer it to clients as a data fabric-as-a-service
ACCESSIBLE
Ensure that an authorized user never loses access to an expected record. Moved records should have seamless pointers, for example. And any protocol used to transmit data should be clearly defined, consistent and whenever possible adhere to a free and open standard.
INTEROPERABLE
An organization can leverage data and metadata across its systems only when its records are structured so users and machines can easily interpret and act on the information. Data should be described by standards-based vocabularies, and references between records or between metadata should be meaningful, specific and consistent.
REUSABLE
Reusability is a natural consequence of designing data structures that promote findability, accessibility and interoperability. By adding descriptions of data quality, licensing and provenance, organizations understand when, how long and in which contexts data is best suited for re-use. “Describing your data well is the key to re-use,” Stoop said.
Explore more from the Data Strategy Acceleration series
unified data fabric
What Your Data Strategy Is Missing
Why Data Access Is Key
Next-Gen Data Architectures