Avoiding Legal Pitfalls Through Savvy Data Governance

A centralized access layer involving data virtualization, in which data remains in sources and is exposed to others as curated data products through a universal semantic interface, is useful for implementing this architecture. Knowledge graphs, facilitated at the departmental level, then connected at a higher level to share data products, provide an alternate take on the same concept. With the semantic layer approach Sengupta described, organizations can dictate policies at a universal level. They can then apply them to data products based on the classifications and tags of the respective data assets comprising them.

A data owner can simply indicate “which data product contains PII (personally identifiable information), a telephone number, a physical address, and the appropriate security policies to govern who can access that, who can see that data, whether it’s redacted or whatever, and it’ s automatically applied,” Moxon explained. This way, organizations facilitate governed sharing of enterprise knowledge, and consumers of that knowledge receive vetted, up-to-date information for their particular applications.

Federated Data Governance

Data marketplaces, in which users search for, compare, preview, and request access to information centrally, extend the utility of the data product notion. When properly executed, a data marketplace gives data owners what Vogt termed a “human-in-the-loop provisioning process” to approve or deny access requests. Such a mechanism solidifies enforcement of data governance policies.

Data marketplaces exemplify the federated data governance paradigm by allowing data owners to implement rules locally for data access and data security. However, by exposing that information through a uniform construct, which provides an experience similar to that of a shopping experience on a retailer’s website (which is how data marketplaces function), they avail themselves of the centralization pivotal to successful data governance. According to Kamien, “Centralizing data is essential to achieve effective governance because it creates an environment that’s unified and controlled where the policies and processes and standards can be applied.”

When such centralization involves a data marketplace, data owners share governed data assets according to the requisite policies and conventions Kamien mentioned. However, they also pick and choose with whom they share that data. Once consumers finish perusing the marketplace to see what knowledge is available for their use cases, they solicit access through what Vogt described as “request work- flows.” According to Vogt, “It could be long-running or temporary access, and that triggers the producers or subject matter experts to approve or deny that request. Then Immuta kicks in to
provision access automatically wherever those data products are.” Some data governance solutions even contain data marketplaces themselves, in addition to integrating with data marketplaces in data catalogs or other tooling.

Monitoring and Traceability

The capacity to monitor access requests in real time, then trace specific requests, approvals, and denials for legal and regulatory demands, is critical for data governance. The comprehensiveness of such insight is fundamental for ensuring that individuals adhere to governance rules. Details such as log file analysis and analysis of specific queries are instrumental in providing this degree of oversight. These techniques are responsible for supplying organizations with “complete visibility and traceability,” Moxon stated. “We’re logging everything so you have a record of who’s accessing the data, when, how, and where they’re physically located in terms of IP address.”