Nine Best Practices for Keeping Bad Actors Out of a Database

CIOs are facing a data security problem that continues to snowball. Cyber-attacks show no signs of stopping, forcing organizations to scramble to figure out how to best protect mission-critical data, particularly when it’s spread across silos in disconnected databases. And it’s not enough to protect from external threats with perimeter security; CIOs also need to handle the increasing number of insider threats. At the same time that they are being pressed to lock down their data, they are also under pressure to drive business innovation by sharing and leveraging their data at a scale never seen before. Locking up the data and never sharing it would bring innovation to a halt. In this eWEEK slide show, based on industry information from Joe Pasqua, executive vice president of products at enterprise NoSQL database provider MarkLogic, we prescribe best practices for securing data without hindering data sharing.

The first step to ensuring the security of your data is knowing what data needs to be secured, MarkLogic Executive Vice President Joe Pasqua said. Organizations need to catalog and understand all the data they have, whether in the cloud or on-premises. This includes everything from internet of things (IoT) data, click streams and other big data to the most sensitive personally identifiable information (PII).

Once an organization knows what data it has in its care, it needs to determine its value—and its sensitivity. Or, to put another way, what would the consequences be if certain data was lost, stolen or misused? Only then can appropriate security and access control measures be put into place that balance risk, consequences and value.

Do the right people have access to the right data for the right amount of time? What if their roles change? What if the nature of the data or governing regulations change? This information is key to appropriately assigning (and, in some cases, revoking) access. It’s also important to ensuring that maximum value can be derived from the data. Yes, you need to keep the wrong people out, but you also need to let the right people in. 

Many database systems can only store a single kind of data. That means data needs to be split up and stored across many different systems. This makes the security problem much harder. Taking a multi-model approach to database security enables organizations to more effectively govern data, managing high-level business concepts from multiple silos and materializing them as entities and relationships. By keeping data and metadata together, details and relationships between data can be effectively tracked over time.

You aren’t starting from scratch. There are strong standards that provide a security benchmark and ensure that all players are on the same page. The Common Criteria Certification, for example, mandates strict access controls and authentication that work with the organization’s existing IT infrastructure. Ensure that your software suppliers are also following these standards.

Anonymization enables enterprises to safely share the right views of data with the right audiences for the right amount of time by removing, replacing or blocking out sensitive information in order to prevent leakage or the violation of laws and regulations. There are different types of anonymization that suit different circumstances. Use a system that provides powerful options and a way to customize the anonymization.

Don’t take an all-or-nothing approach. This limits the usefulness of your data by restricting it based on the most sensitive data it contains. For example, an employee record (or document) may have lots of data that is visible by several different roles but some that is much more sensitive. If access is determined at the record level, then access to the entire record would be blocked because of just one or two sensitive items. Use a system that allows you to specify, in a very flexible, very granular way, which items are visible to which roles. Once again, this enables more sharing with less risk.

Encrypting your data at rest is a must these days, especially as organizations move to the cloud. One of the most crucial parts of any encryption system is how it manages keys, and the best practice is to use an external, third-party key management system (KMS) that is deployed and managed independently from other systems. An external KMS securely stores keys and provides them on demand to authorized systems. This separation of concerns enables an additional level of security and consistency across all systems performing encryption.   

To protect against data breaches and insider threats while at the same time having the ability to easily and securely share information, organizations must implement strong, flexible and comprehensive data policies across the entire organization. These policies are often based on key metadata about the information being stored such as: where the data came from, who touched it last, how it can be used, whether it is bound to a regulation, and which employees are allowed to see it and for how long. It is impossible to implement business-level policies without knowing these sorts of facts about your data. Most database systems don’t do this, which puts the burden back on the organization.