Microsoft SQL Server 2012 Database Security Technical Implementation Guide

This control is intended to address the confidentiality and integrity of information at rest in non-mobile devices and covers user information and system information. Information at rest refers to the state of information when it is located on a secondary storage device (e.g., disk drive, tape drive) within an organizational information system. Applications and application users generate information throughout the course of their application use. User-generated data, as well as, application-specific configuration data, needs to be protected. Configurations and/or rule sets for firewalls, gateways, intrusion detection/prevention systems, filtering routers, and authenticator content are examples of system information likely requiring protection. Organizations may choose to employ different mechanisms to achieve confidentiality and integrity protections, as appropriate. If the confidentiality and integrity of SQL Server data is not protected, the data will be open to compromise and unauthorized modification. Protective measures include encryption, physical security of the facility where the storage devices reside, operating system file permissions, and organizational controls. Each of these should be applied as necessary and appropriate.

Security attributes are abstractions representing the basic properties or characteristics of an entity (e.g., subjects and objects) with respect to safeguarding information. These attributes are typically associated with internal data structures (e.g., records, buffers, files) within the information system and are used to enable the implementation of access control and flow control policies; reflect special dissemination, handling, or distribution instructions; or support other aspects of the information security policy. One example includes marking data as classified or FOUO. These security attributes may be assigned manually or during data processing but, either way, it is imperative these assignments are maintained while the data is in storage. If the security attributes are lost when the data is stored, there is the risk of a data compromise. The sensitivity marking or labeling of stored data items promotes the correct handling and protection of data. Without such notification, the user may unwittingly disclose sensitive data to unauthorized users. (Earlier releases of this STIG suggested using the SQL Server Label Security Toolkit, from codeplex.com. However, codeplex.com has been shut down, and it is unclear whether the Toolkit is still supported. If the organization does have access to the Toolkit, it may still be used, provided the organization accepts responsibility for its support.)

Security attributes are abstractions representing the basic properties or characteristics of an entity (e.g., subjects and objects) with respect to safeguarding information. These attributes are typically associated with internal data structures (e.g., records, buffers, files) within the application and are used to enable the implementation of access control and flow control policies; reflect special dissemination, handling, or distribution instructions; or support other aspects of the information security policy. Organizations define the security attributes of their data (e.g., classified, FOUO). Applications generating and/or processing data assigned these organization-defined security attributes must maintain the binding of these attributes to the data when the data is transmitted. If the application does not maintain the data security attributes when it transmits the data, there is a risk of data compromise. The sensitivity marking or labeling of data items promotes the correct handling and protection of data. Without such notification, the user may unwittingly disclose sensitive data to unauthorized users. Security labels must be correctly maintained throughout transmission. (Earlier releases of this STIG suggested using the SQL Server Label Security Toolkit, from codeplex.com. However, codeplex.com has been shut down, and it is unclear whether the Toolkit is still supported. If the organization does have access to the Toolkit, it may still be used, provided the organization accepts responsibility for its support.)

Security attributes are abstractions representing the basic properties or characteristics of an entity (e.g., subjects and objects) with respect to safeguarding information. These attributes are typically associated with internal data structures (e.g., records, buffers, files) within the application and are used to enable the implementation of access control and flow control policies; reflect special dissemination, handling, or distribution instructions; or support other aspects of the information security policy. Organizations define the security attributes of their data (e.g., classified, FOUO). Applications generating and/or processing data assigned these organization-defined security attributes must maintain the binding of these attributes to the data when the data is transmitted. If the application does not maintain the data security attributes when it transmits the data, there is a risk of data compromise. The sensitivity marking or labeling of data items promotes the correct handling and protection of data. Without such notification, the user may unwittingly disclose sensitive data to unauthorized users. Security labels must be correctly maintained throughout transmission. (Earlier releases of this STIG suggested using the SQL Server Label Security Toolkit, from codeplex.com. However, codeplex.com has been shut down, and it is unclear whether the Toolkit is still supported. If the organization does have access to the Toolkit, it may still be used, provided the organization accepts responsibility for its support.)

Security attributes are abstractions representing the basic properties or characteristics of an entity (e.g., subjects and objects) with respect to safeguarding information. These attributes are typically associated with internal data structures (e.g., records, buffers, files) within the information system and are used to enable the implementation of access control and flow control policies; reflect special dissemination, handling, or distribution instructions, or support other aspects of the information security policy. Examples of application security attributes are classified, FOUO, sensitive, etc. Throughout the course of normal usage, authorized users of applications that handle sensitive data will have the need to associate security attributes with information. Applications that maintain the binding of organization-defined security attributes to data must ensure authorized users can associate security attributes with information. For databases, this is accomplished via labeling.

Discretionary Access Control (DAC) is based on the premise that individual users are "owners" of objects and therefore have discretion over who should be authorized to access the object and in which mode (e.g., read or write). Ownership is usually acquired as a consequence of creating the object or via specified ownership assignment. DAC allows the owner to determine who will have access to objects they control. An example of DAC includes user-controlled file permissions. DAC models have the potential for the access controls to propagate without limit, resulting in unauthorized access to said objects. When applications provide a discretionary access control mechanism, the application must be able to limit the propagation of those access rights. The DBMS must ensure the recipient of object permissions possesses only the access intended. The database must enforce the ability to limit unauthorized rights propagation. If propagation is not prevented, users can continue to grant rights to other users without limit.true

Applications employ the concept of least privilege for specific duties and information systems (including specific functions, ports, protocols, and services). The concept of least privilege is also applied to information system processes, ensuring that the processes operate at privilege levels no higher than necessary to accomplish required organizational missions and/or functions. Organizations consider the creation of additional processes, roles, and information system accounts as necessary to achieve least privilege. Organizations also apply least privilege concepts to the design, development, implementation, and operations of information systems. Developers granted elevated database and/or operating system privileges on production databases can affect the operation and/or security of the database system. Operating system and database privileges assigned to developers on production systems should not be allowed.

Applications employ the concept of least privilege for specific duties and information systems (including specific functions, ports, protocols, and services). The concept of least privilege is also applied to information system processes, ensuring that the processes operate at privilege levels no higher than necessary to accomplish required organizational missions and/or functions. Organizations consider the creation of additional processes, roles, and information system accounts as necessary to achieve least privilege. Organizations also apply least privilege concepts to the design, development, implementation, and operations of information systems. Developers granted elevated database and/or operating system privileges on systems that support both development and production databases can affect the operation and/or security of the production database system. Operating system and database privileges assigned to developers on shared development and production systems must be restricted.

SQL Server must employ the concept of least privilege for specific duties and information systems (including specific functions, ports, protocols, and services). The concept of least privilege is also applied to information system processes, ensuring that the processes operate at privilege levels no higher than necessary to accomplish required organizational missions and/or functions. Organizations consider the creation of additional processes, roles, and information system accounts as necessary to achieve least privilege. Organizations also apply least privilege concepts to the design, development, implementation, and operations of information systems. Privileges granted outside the role of the application user job function are more likely to go unmanaged or without oversight for authorization. Maintenance of privileges using roles defined for discrete job functions offers improved oversight of application user privilege assignments and helps to protect against unauthorized privilege assignment. SQL Server built-in administrative privileges, built-in server roles and built-in database roles must not be assigned directly to administrative user accounts (that is, server logins and database users). If administrative user accounts have direct access to administrative roles, this access must be removed, with the exception of administrative roles that the DBMS assigns to the special database principal [dbo], and will not allow to be altered. The built-in server role "sysadmin" is a partial exception. This cannot be granted to a user-defined role, only to a login account. Most (not necessarily all) database administrators will need to be members of sysadmin. Without this, most DBCC commands and the system stored procedures/functions listed below are unavailable. The users who require such access must be documented and approved.true

When dealing with unauthorized SQL Server job submissions, it should be noted any unauthorized job submissions to SQL Server job/batch queues can potentially have significant effects on the overall security of the system. If SQL Server were to allow any user to make SQL Server job/batch queue submissions, then those submissions might lead to a compromise of system integrity and/or data. This requirement is contingent upon the SQL Server job/batch queue being review regularly for unauthorized submissions. Accordingly, only qualified and authorized individuals shall be allowed to obtain access to submit SQL Server jobs. Job/batch queue submissions must adhere to an organization-defined job submission process. Unmanaged changes that occur to SQL Server job/batch queues can lead to a compromised system.

Audit records can be generated from various components within the information system (e.g., network interface, hard disk, modem, etc.). From an application perspective, certain specific application functionalities may be audited as well. The list of audited events is the set of events for which audits are to be generated. This set of events is typically a subset of the list of all events for which the system is capable of generating audit records. Examples are auditable events, time stamps, source and destination addresses, user/process identifiers, event descriptions, success/fail indications, file names involved, and access control or flow control rules invoked. Organizations define which application components shall provide auditable events. The DBMS must provide auditing for the list of events defined by the organization or risk negatively impacting forensic investigations into malicious behavior in the information system.

When dealing with change control issues, it should be noted, any changes to the hardware, software, and/or firmware components of SQL Server and/or application can potentially have significant effects on the overall security of the system. If SQL Server were to allow any user to make changes to software libraries, then those changes might be implemented without undergoing the appropriate testing and approvals that are part of a robust change management process. This requirement is contingent upon the language in which the application is programmed, as many application architectures in use today incorporate their software libraries into, and make them inseparable from, their compiled distributions, rendering them static and version-dependent. However, this requirement does apply to applications with software libraries accessible and configurable, as in the case of interpreted languages. Accordingly, only qualified and authorized individuals shall be allowed to obtain access to SQL Server components for purposes of initiating changes, including upgrades and modifications. Unmanaged changes that occur to the SQL Server software libraries or configuration, such as Functions, can lead to unauthorized or compromised installations.true

When dealing with change control issues, it should be noted, any changes to the hardware, software, and/or firmware components of SQL Server and/or application can potentially have significant effects on the overall security of the system. If SQL Server were to allow any user to make changes to software libraries, then those changes might be implemented without undergoing the appropriate testing and approvals that are part of a robust change management process. This requirement is contingent upon the language in which the application is programmed, as many application architectures in use today incorporate their software libraries into, and make them inseparable from, their compiled distributions, rendering them static and version-dependent. However, this requirement does apply to applications with software libraries accessible and configurable, as in the case of interpreted languages. Accordingly, only qualified and authorized individuals shall be allowed to obtain access to SQL Server components for purposes of initiating changes, including upgrades and modifications. Unmanaged changes that occur to the SQL Server software libraries or configuration, such as Triggers, can lead to unauthorized or compromised installations.

When dealing with change control issues, it should be noted, any changes to the hardware, software, and/or firmware components of SQL Server and/or application can potentially have significant effects on the overall security of the system. If SQL Server were to allow any user to make changes to software libraries, then those changes might be implemented without undergoing the appropriate testing and approvals that are part of a robust change management process. This requirement is contingent upon the language in which the application is programmed, as many application architectures in use today incorporate their software libraries into, and make them inseparable from, their compiled distributions, rendering them static and version-dependent. However, this requirement does apply to applications with software libraries accessible and configurable, as in the case of interpreted languages. Accordingly, only qualified and authorized individuals shall be allowed to obtain access to SQL Server components for purposes of initiating changes, including upgrades and modifications. Unmanaged changes that occur to the SQL Server software libraries or configuration, such as Stored Procedures, can lead to unauthorized or compromised installations.

SQL Server database ownership is a higher level privilege that grants full rights to everything in that database, including the right to grant privileges to others. SQL Server requires that the owner of a database object be a user, and only one user can be the assigned owner of a database object. This tends to minimize the risk that multiple users could gain unauthorized access, except the one individual who is the owner. Within the database, object ownership implies full privileges to the owned object, including the privilege to assign access to the owned objects to other subjects. Unmanaged or uncontrolled ownership of databases can lead to unauthorized granting of privileges and database alterations.

Information systems are capable of providing a wide variety of functions and services. Some of the functions and services, provided by default, may not be necessary to support essential organizational operations (e.g., key missions, functions). It is detrimental for applications to provide or install by default, functionality exceeding requirements or mission objectives. Examples include, but are not limited to, installing advertising software demonstrations or browser plug-ins not related to requirements or providing a wide array of functionality not required for every mission, yet cannot be disabled. Applications must adhere to the principles of least functionality by providing only essential capabilities. Unused and unnecessary SQL Server components increase the number of available attack vectors to SQL Server by introducing additional targets for attack. By minimizing the services and applications installed on the system, the number of potential vulnerabilities is reduced.

When data is written to digital media, such as hard drives, mobile computers, external/removable hard drives, personal digital assistants, flash/thumb drives, etc., there is risk of data loss and/or compromise. An organizational assessment of risk guides the selection of media and associated information contained on that media requiring restricted access. Organizations need to document, in policy and procedures, the media requiring restricted access, individuals authorized to access the media, and the specific measures taken to restrict access. Fewer protection measures are needed for media containing information determined by the organization to be in the public domain, to be publicly releasable, or to have limited or no adverse impact if accessed by other than authorized personnel. In these situations, it is assumed the physical access controls where the media resides provide adequate protection. As part of a defense-in-depth strategy, the organization considers routinely encrypting information at rest on selected secondary storage devices. The decision whether to employ cryptography is the responsibility of the information owner/steward, who exercises discretion within the framework of applicable rules, policies and law. The selection of the cryptographic mechanisms used is based upon maintaining the confidentiality and integrity of the information. The strength of mechanisms is commensurate with the classification and sensitivity of the information. Information at rest, when not encrypted, is open to compromise from attackers who have gained unauthorized access to the data files.

Cryptography is only as strong as the encryption modules/algorithms employed to encrypt the data. Use of weak or untested encryption algorithms undermines the purposes of utilizing encryption to protect data. Use of cryptography to provide confidentiality and non-repudiation is not effective unless strong methods are employed with its use. Many earlier encryption methods and modules have been broken and/or overtaken by increasing computing power. The NIST FIPS 140-2 cryptographic standards provide proven methods and strengths to employ cryptography effectively. Detailed information on the NIST Cryptographic Module Validation Program (CMVP) is available at the following website: http://csrc.nist.gov/groups/STM/cmvp/index.html.

When not encrypted by the Service Master Key, system administrators or application administrators may access and use the Database Master Key to view sensitive data that they are not authorized to view. Where alternate encryption means are not feasible, encryption by the Service Master Key may be necessary. To help protect sensitive data from unauthorized access by DBAs, mitigations may be in order. Mitigations may include automatic alerts or other audit events when the Database Master Key is accessed outside of the application or by a DBA account.true

Storage of the Database Master Key password in a database credential allows decryption of sensitive data by privileged users who may not have a need-to-know requirement to access the data.

Data within the database is protected by use of encryption. The symmetric keys are critical for this process. If the symmetric keys were to be compromised the data could be disclosed to unauthorized personnel.true

This control is intended to address the confidentiality and integrity of information at rest in non-mobile devices and covers user information and system information. If the data is not encrypted, it is subject to compromise and unauthorized disclosure. Note: the system databases (master, msdb, model, resource and tempdb) cannot be encrypted. The decision whether to employ cryptography is the responsibility of the information owner/steward, who exercises discretion within the framework of applicable rules, policies and law.

The purpose of this control is to prevent information, including encrypted representations of information, produced by the actions of a prior user/role (or the actions of a process acting on behalf of a prior user/role) from being available to any current user/role (or current process) that obtains access to a shared system resource (e.g., registers, main memory, secondary storage) after the resource has been released back to the information system. Control of information in shared resources is also referred to as object reuse. Data used for the development and testing of applications often involves copying data from production. It is important that specific procedures exist for this process, so copies of sensitive data are not misplaced or left in a temporary location without the proper controls.

Application management includes the ability to control the number of users and user sessions utilizing an application. Limiting the number of allowed users, and sessions per user, is helpful in limiting risks related to DoS attacks. This requirement addresses concurrent session control for a single information system account and does not address concurrent sessions by a single user via multiple system accounts. This requirement may be met via the application or by utilizing information system session control provided by a web server with specialized session management capabilities. If it has been specified that this requirement will be handled by the application, the capability to limit the maximum number of concurrent single user sessions must be designed and built into the application. The organization will need to define the maximum number of concurrent sessions for SQL Server accounts by account type, by account, or a combination thereof and SQL Server shall enforce this requirement. Unlimited concurrent connections to SQL Server could allow a successful DoS attack by exhausting connection resources.

Invalid user input occurs when a user inserts data or characters into an application’s data entry fields and the application is unprepared to process that data. This results in unanticipated application behavior potentially leading to an application or information system compromise. Invalid user input is one of the primary methods employed when attempting to compromise an application. SQL Server needs to validate the data user’s attempt to input to the application for processing. Rules for checking the valid syntax and semantics of information system inputs (e.g., character set, length, numerical range, acceptable values) are in place to verify inputs match specified definitions for format and content. Inputs passed to interpreters are prescreened to prevent the content from being unintentionally interpreted as commands. A poorly designed database system can have many problems. A common issue with these types of systems is the missed opportunity to use constraints. While this matter is of great importance to the secure operation of database management systems, the DBA in a typical installation will communicate with the application development/support staff to obtain assurance that this requirement is met.

SQL Server's fixed (built-in) server roles, especially [sysadmin], have powerful capabilities that could cause great harm if misused, so their use must be tightly controlled. The SQL Server instance uses each database's TRUSTWORTHY property to guard against tampering that could enable unwarranted privilege escalation. When TRUSTWORTHY is 0/False/Off, SQL Server prevents the database from accessing resources in other databases. When TRUSTWORTHY is 1/True/On, SQL Server permits access to other databases (subject to other protections). SQL Server sets TRUSTWORTHY OFF when it creates a new database. SQL Server forces TRUSTWORTHY OFF, irrespective of its prior value, when an existing database is attached to it, to address the possibility that an adversary may have tampered with the database, introducing malicious code. To set TRUSTWORTHY ON, an account with the [sysadmin] role must issue an ALTER DATABASE command. Although SQL Server itself treats this property conservatively, application installer programs may set TRUSTWORTHY ON and leave it on. This provides an opportunity for misuse. When TRUSTWORTHY is ON, users of the database can take advantage of the database owner's privileges, by impersonating the owner. This can have particularly serious consequences if the database owner is the [sa] login (which may have been renamed in accordance with SQL2-00-010200, and disabled in accordance with SQL2-00-017100, but nonetheless can be invoked in an EXECUTE AS USER = 'dbo' statement, or CREATE PROCEDURE ... WITH EXECUTE AS OWNER ...). The [sa] login cannot be removed from the [sysadmin] role. The user impersonating [sa] - or another [sysadmin] account - is then able to perform administrative actions across all databases under the instance, including making any himself or any other login a member of [sysadmin]. Most of the other fixed server roles could be similarly abused. Therefore, TRUSTWORTHY must not be used on databases owned by logins that are members of the fixed server roles. Further, if TRUSTWORTHY is to be used for any other database, the need must be documented and approved. The system database [msdb] is an exception: it is required to be TRUSTWORTHY.

SQL Server's fixed (built-in) server roles, especially [sysadmin], have powerful capabilities that could cause great harm if misused, so their use must be tightly controlled. The SQL Server instance uses each database's TRUSTWORTHY property to guard against tampering that could enable unwarranted privilege escalation. When TRUSTWORTHY is 0/False/Off, SQL Server prevents the database from accessing resources in other databases. When TRUSTWORTHY is 1/True/On, SQL Server permits access to other databases (subject to other protections). SQL Server sets TRUSTWORTHY OFF when it creates a new database. SQL Server forces TRUSTWORTHY OFF, irrespective of its prior value, when an existing database is attached to it, to address the possibility that an adversary may have tampered with the database, introducing malicious code. To set TRUSTWORTHY ON, an account with the [sysadmin] role must issue an ALTER DATABASE command. Although SQL Server itself treats this property conservatively, application installer programs may set TRUSTWORTHY ON and leave it on. This provides an opportunity for misuse. When TRUSTWORTHY is ON, users of the database can take advantage of the database owner's privileges, by impersonating the owner. This can have particularly serious consequences if the database owner is the [sa] login (which may have been renamed in accordance with SQL2-00-010200, and disabled in accordance with SQL2-00-017100, but nonetheless can be invoked in an EXECUTE AS USER = 'dbo' statement, or CREATE PROCEDURE ... WITH EXECUTE AS OWNER ...). The [sa] login cannot be removed from the [sysadmin] role. The user impersonating [sa] - or another [sysadmin] account - is then able to perform administrative actions across all databases under the instance, including making any himself or any other login a member of [sysadmin]. Most of the other fixed server roles could be similarly abused. Therefore, TRUSTWORTHY must not be used on databases owned by logins that are members of the fixed server roles. Further, if TRUSTWORTHY is to be used for any other database, the need must be documented and approved. The system database [msdb] is an exception: it is required to be TRUSTWORTHY.