OBSTACLES
What are the obstacles to fully embracing data fabric or data mesh? Are enterprises ready? Both data mesh and data fabric have their challenges.
“Data meshes can require significant investment in terms of time, resources, and expertise while being complex to operate and maintain,” said Sarkar. “There needs to be a clear definition and common understanding regarding establishing the concept of data domain, domain owners, and the data governance model.”
In addition, “data mesh also requires a well-defined process for automatically updating data catalogs,” Sarkar continued. “For many organizations, implementing a good data catalog strategy has been a challenge and requires a constant and dedicated effort to implement and maintain.”
Similarly, data fabric “also requires a strong data governance and data catalog strategy,” Sarkar added. “Setting up and managing data fabric can be complex and requires a high level of expertise. It can be challenging for organizations with limited IT and data skill sets.”
Data fabric architectures are often met with “resistance to relinquishing control of the underlying first layer: physical connections,” said Mallory. “This new way of consuming and managing networks might pose an adjustment for some IT departments, but the opportunities with fabric-based connections will accelerate digital transformation initiatives across industries.”
Talent shortages and change management also can be challenging in data mesh implementations. Deep domain knowledge and specialized training “are key to an organization’s readiness to embrace data fabric or data mesh,” Ashton said. Still, “data mesh is a hard architecture to implement because it goes against a lot of legacy structures and processes,” said Kupriyanova. This is something enterprises may not be ready for. “Because data mesh requires effective governance, the architecture becomes even more complex to implement. Data governance is one of the hardest processes to implement in any organization and is the first to go when there are financial pressures.”
Data mesh “is a distinctly different approach that requires a certain level of maturity to deploy, specifically in terms of governance,” Kupriyanova continued. “It’s challenging to implement for clients that need it most.”
The hyper-distributed environments associated with mesh mean greater demands for data engineering, data science, and analytic functional management talent across all domains. While centralizing talent teams and governance may address these issues, “a centralized team frequently becomes a bottleneck, and low priority functions rarely got the attention they need,” Dow noted.
While data mesh promises decentralization and flexibility, “some implementations can go too far, tossing out the crucial benefits of centralized coordination and cross-functional agility,” Ashton warned. “They’re essentially tossing the baby out with the bathwater, sacrificing valuable capabilities in the name of reduced complexity.”
The bottom line is that every organization is different—with “unique goals, data characteristics, technical expertise, and business requirements,” Thompson said. “There is no one-size-fits-all solution. But selecting a framework that prioritizes the organization’s business context ensures that the chosen architecture integrates smoothly with its objectives.”