Data security in the cloud is of chief concern not only to healthcare and financial services, but anyone with sensitive data of any kind that should only be disclosed to authorized parties. No discussion of enterprise security would be complete without a look at data protection and governance.
For purposes of this discussion, data comes in two forms:
- Structured. Structured data refers to kinds of data with a high level of organization, such as information stored in a relational database, as in Microsoft SQL Server.
- Unstructured. Unstructured data refers to data that is not contained in a database or some other type of data structure. Examples include email messages, Word documents, PowerPoint presentations and instant messages.
Important considerations in data protection and governance include data classification and rights management, encryption at-rest and in-flight, as well as management and storage of encryption keys and other secrets related to securing data.
Securing Structured Data In-Flight & In Use
SQL Server 2016 (both SQL in VMs and Azure SQL) introduces some new capabilities to prevent unintentional leakage of data by misconfigured applications or security controls. Key highlights are listed below:
#1 Always Encrypted:
This is a client-side encryption capability, enabling the application to encrypt data so the SQL server (or service if using Azure SQL) can never see the data. This is particularly useful for protecting content such as SIN/SSN, Credit Card, and private health identifiers.
#2 Row-Level Security:
This allows the organization to create policies which only return data rows appropriate for the user executing the query. For example, this allows a hospital to only return health information of patients directly related to a nurse, or a bank teller to only see rows returned which are relevant to their role. For more info, see https://msdn.microsoft.com/en-us/library/dn765131.aspx.
#3 Dynamic Data Masking:
This allows the organization to create policies to mask data in a particular field. For example, an agent at a call center may identify callers by the last few digits of their social security number or credit card number, but those pieces of information should not be fully exposed to the agent. Dynamic Data Masking can be configured on the SQL server to return the application query for the credit card numbers as XXXX-XXXX-XXXX-1234.
These capabilities help prevent and mitigate accidental exposure of data while it is in-flight or in-use by a front-end application. For more info, see https://msdn.microsoft.com/en-us/library/mt130841.aspx.
Securing Structured Data At-Rest
Protection of SQL data at-rest is a feature that has been around for a long time now, which the SQL Server product team at Microsoft has enhanced in the 2016 release.
#4 SQL Transparent Data Encryption
In order to protect structured data at-rest, Microsoft first introduced SQL Transparent Data Encryption in SQL Server 2008. This technology protects data by performing I/O encryption for SQL database and log files. Traditionally a certificate that SQL Server manages (and is stored locally within the SQL master database) would protect this data encryption key (DEK). In June 2016, Microsoft made a significant enhancement to this capability by making generally available a SQL Server Connector for Azure Key Vault.
Image credit: Microsoft
This allows organizations to separate SQL and Security Administrator roles, enabling a SQL Administrator to leverage a key managed by the security operators in Azure Key Vault, with a full audit trail should the SQL administrator turn rogue. This connector can also be used for encrypting specific database columns and backups, and is backward compatible all the way back to SQL 2008.
Detecting SQL Threats
In addition to securing SQL data, we also need to consider protecting data sources from the threats that would lead to breach.
#5 SQL Threat Detection
Running SQL in the cloud brings some additional benefits. For databases running on the Azure SQL service, the new SQL Threat Detection service monitors database activity and access, building profiles to identify anomalous behavior or access. If suspicious activity is detected, security personnel can get immediate notification about the activities as they occur. Each notification provides details of the suspicious activity and recommendations on remediating the threat.
SQL Threat Detection for Azure SQL Database can detect threats such as the following:
- Potential Vulnerabilities: SQL Threat Detection will detect common misconfigurations in application connectivity to the SQL data, and provide recommendations to the administrators to harden the environment.
- SQL Injection Attacks: One of the most common approaches to data extraction is to insert a SQL query into an unprotected web form, causing the form to return data that was unintended. SQL Threat Detection can identify if an attacker is attempting to leverage this mechanism to extract data.
- Anomalous Database Access: If a compromised database administrator account starts to execute queries from an abnormal location, SQL Threat Detection can detect and alert on the potential insider threat or identity compromise, enabling the security personnel to update firewall rules or disable the account.
SQL Threat Detection for Azure SQL Database is a powerful new tool in detecting potential data leakage threats. For more info, see https://docs.microsoft.com/en-us/azure/sql-database/sql-database-threat-detection.
I hope you’ve found this short read on some of Microsoft’s capabilities for protecting structured data valuable. Questions or comments? Feel free to leave your thoughts in the comments section at the end of this article.