Enterprise data security is paramount when migrating data to the cloud. Previously, with on-prem environments, controls were contained within a finite and well-defined border, making it simpler to manage data privacy and security. The cloud has expanded the reach of enterprise businesses, but it has also opened the security perimeter exposing enterprise data to greater risks. As a result, many data governance teams are not permitting migration to the cloud without securing the data first.
Encryption is the key to secure cloud migration and is the primary tool for ensuring compliance to privacy regulations worldwide including GDPR, CCPA, and most recently, LGPD. All of these regulations require organizations to manage the privacy of customer data and are upending the way enterprises manage and share data as individuals have more rights over their personal data and how it can be used.
Understanding Two Critical Prerequisites to Effective Data Encryption
While important, encrypting data is only part of the story as there are two prerequisites to data encryption that must be addressed beforehand. The first is to understand the enterprise data landscape. The detection and identification of regulated and personally identifiable information (PII) such as name, address and credit/social security numbers are critical because if you can’t locate and identify it, you certainly can’t secure it.
The second is to understand the data itself and how it will be utilized, as there needs to be a balance between security and usability/accessibility. Data that is completely inaccessible is 100% secure but completely useless for the depth of data science and analytics needed to drive the “next best decision.”
Selecting Encryption Options Depends on Requirements
Once the prerequisites have been completed, the next step is to determine what type of encryption makes sense. Ask things like: Do the regulatory, or internal data governance regulations require the data to be completely obscured to the audience? Or only partially, like a typical credit card statement where only the last four digits are visible on the statement, and the rest are represented by xxxx-xxx-1234?
There are other situations where the same data requires different types of encryption, depending on the intended audience or user of the data. For example, an HR team may need a social security number to be clear and unencrypted to request a credit report for a potential new hire. However, other audiences would only require partial encryption, such as the finance team needing only the last four numbers of a credit card number to confirm a transaction. Or a data analytics team that only needs to know if the data represents a unique individual or not. Other privacy regulations require data to be completely encrypted so that data cannot but viewed or reverse engineered to reveal any PII whatsoever.
To understand the required type of encryption, security leaders need to work with data teams to understand the regulatory requirements and use cases. Balancing this dual mandate of making data accessible for analytics without compromising security/governance requirements is a joint responsibility shared between data governance, data security, data analytics and IT (security) teams.
These groups have different options - such as masking, anonymization, pseudo anonymization, and encryption, each providing a different outcome. These options are defined as:
- Masking: the process of ensuring original data is obfuscated by modified content but still remains usable.
- Anonymization: a method of ensuring privacy data such as individual names and contact information are inaccessible and irreversible from the data itself, a requirement for most compliance regulations such as GDPR, CCPA and LGPD.
- Pseudo anonymization: the replacement of PII with a set of artificial identifiers. Pseudonymized data can be restored to its original state with the addition of information which then allows individuals to be re-identified, while anonymized data can never be restored to its original state.
- Encryption: encoding original data into an alternative format, such that only authorized parties (i.e. those with the encryption key) can decrypt and access the original data.
Encryption Considerations
The ability to analyze data relies on its ability to be encrypted, which is why it’s important to select the right tool for the regulatory requirements and the intended use case. This becomes more difficult to manage when migrating data to the cloud, due to the diversity of public cloud services, and the different regulations in different geographies. Mapping data in the cloud and managing access control policies with the right kind of encryption can ensure consistent compliance policies including GDPR, CCPA, HIPAA and other regulations for data distributed across multiple cloud databases, analytics platforms, reporting systems, and geographies. More importantly, it is the best way for IT and security teams to ensure data security and privacy without hindering data analytics.