Service Principal Name (SPN) Support in Client Connections
Applies To: SQL Server 2016 Preview
Beginning with SQL Server 2008, support for service principal names (SPNs) has been extended to enable mutual authentication across all protocols. In previous versions of SQL Server, SPNs were only supported for Kerberos over TCP when the default SPN for the SQL Server instance was registered with Active Directory.
SPNs are used by the authentication protocol to determine the account in which a SQL Server instance runs. If the instance account is known, Kerberos authentication can be used to provide mutual authentication by the client and server. If the instance account is not known, NTLM authentication, which only provides authentication of the client by the server, is used. Currently, SQL Server Native Client performs the authentication lookup, deriving the SPN from the instance name and network connection properties. SQL Server instances will attempt to register SPNs on startup, or they can be registered manually. However, registration will fail if there are insufficient access rights for the account that attempts to register the SPNs.
Domain and computer accounts are registered automatically in Active Directory. These can be used as SPNs, or administators can define their own SPNs. SQL Server makes secure authentication more manageable and reliable by allowing clients to directly specify the SPN to use.
An SPN specified by a client application is only used when a connection is made with Windows integrated security.
Microsoft Kerberos Configuration Manager for SQL Server is a diagnostic tool that helps troubleshoot Kerberos related connectivity issues with SQL Server. For more information, see Microsoft Kerberos Configuration Manager for SQL Server.
For more information about Kerberos, see the following articles:
The following table describes the most common scenarios in which client applications can enable secure authentication.
A legacy application does not specify an SPN.
This compatibility scenario guarantees that there will be no behavior change for applications developed for previous versions of SQL Server. If no SPN is specified, the application relies on generated SPNs and has no knowledge of which authentication method is used.
A client application using the current version of SQL Server Native Client specifies an SPN in the connection string as a domain user or computer account, as an instance-specific SPN, or as a user-defined string.
The ServerSPN keyword can be used in a provider, initialization, or connection string to do the following:
-Specify the account used by the SQL Server instance for a connection. This simplifies access to Kerberos authentication. If a Kerberos Key Distribution Center (KDC) is present and the correct account is specified, Kerberos authentication is more likely to be used than NTLM. The KDC usually resides on the same computer as the domain controller.
-Specify an SPN to look up the service account for the SQL Server instance. For every SQL Server instance, two default SPNs are generated that can be used for this purpose. These keys are not guaranteed to be present in Active Directory, however, so in this situation Kerberos authentication is not guaranteed.
-Specify an SPN that will be used to look up the service account for the SQL Server instance. This can be any user-defined string that maps to the service account. In this case, the key must be registered manually in the KDC and must satisfy the rules for a user-defined SPN.
The FailoverPartnerSPN keyword can be used to specify the SPN for the failover partner server. The range of account and Active Directory key values is the same as the values you can specify for the principal server.
An ODBC application specifies an SPN as a connection attribute for the principal server or failover partner server.
The connection attribute SQL_COPT_SS_SERVER_SPN can be used to specify the SPN for a connection to the principal server.
The connection attribute SQL_COPT_SS_FAILOVER_PARTNER_SPN can be used to specify the SPN for the failover partner server.
An OLE DB application specifies an SPN as a data source initialization property for the principal server or for a failover partner server.
The connection property SSPROP_INIT_SERVER_SPN in the DBPROPSET_SQLSERVERDBINIT property set can be used to specify the SPN for a connection.
The connection property SSPROP_INIT_FAILOVER_PARTNER_SPN in the DBPROPSET_SQLSERVERDBINIT can be used to specify the SPN for the failover partner server.
A user specifies an SPN for a server or failover partner server in an ODBC data source name (DSN).
The SPN can be specified in an ODBC DSN through the DSN setup dialog boxes.
The user specifies an SPN for a server or failover partner server in an OLE DB Data Link or Login dialog box.
The SPN can be specified in a Data Link or Login dialog box. The Login dialog box can be used with either ODBC or OLE DB.
An ODBC application determines the authentication method used to establish a connection.
When a connection has been opened successfully, an application can query the connection attribute SQL_COPT_SS_INTEGRATED_AUTHENTICATION_METHOD to determine which authentication method was used. The values will include, but are not limited to, NTLM and Kerberos.
An OLE DB application determines the authentication method used to establish a connection.
When a connection has been opened successfully, an application can query the connection property SSPROP_AUTHENTICATION_METHOD in the DBPROPSET_SQLSERVERDATASOURCEINFO property set to determine which authentication method was used. The values will include, but are not limited to, NTLM and Kerberos.
SPNs are not stored in the failover cache and therefore cannot be passed between connections. SPNs will be used on all connection attempts to the principal and partner when specified in the connection string or connection attributes.
Applications should be aware that specifying SPNs in some but not all connection strings can contribute to pool fragmentation.
Applications can programmatically specify SPNs as connection attributes, instead of specifying connection string keywords. This can help manage connection pool fragmentation.
Applications should be aware that SPNs in connection strings can be overridden by setting the corresponding connection attributes, but connection strings used by connection pooling will use the connection string values for pooling purposes.
The new connection behavior is implemented by the client; therefore, it is not specific to a version of SQL Server.
When linked servers are created, the @provstr parameter of sp_addlinkedserver can be used to specify the server and failover partner SPNs. The benefits of doing this are the same as specifiying SPNs in client connection strings: It is simpler and more reliable to establish connections that use Kerberos authentication.
Delegation with linked servers requires Kerberos authentication.
When choosing whether to specify SPNs in an application (through connection strings) or programmatically through connection properties (rather than relying on the defalt provider generated SPNs), consider the following factors:
Security: Does the specified SPN disclose information that is protected?
Reliability: To enable the use of default SPNs, the service account in which the SQL Server instance runs must have sufficient privileges to update the Active Directory on the KDC.
Convenience and location transparency: How will an application's SPNs be affected if its database moves to a different SQL Server instance? This applies to both the principal server and its failover partner if you use database mirroring. If a server change means that SPNs must be changed, how will this affect applications? Will any changes be managed?
You can specify an SPN in dialog boxes and in code. This section discusses how you can specify an SPN.
The maximum length for an SPN is 260 characters.
The syntax that SPNs use in connection string or connection attributes is as follows:
The provider-generated, default SPN for a default instance when a protocol other than TCP is used.
fqdn is a fully-qualified domain name.
The provider-generated, default SPN when TCP is used.
port is a TCP port number.
The provider-generated, default SPN for a named instance when a protocol other than TCP is used.
InstanceName is a SQL Server instance name.
The SPN that maps to built-in computer accounts that are automatically registered by Windows.
Direct specification of a domain account.
Username is a Windows user account name.
Domain is a Windows domain name or fully-qualified domain name.
Direct specification of a computer account.
(If the server you are connecting to is running under LOCAL SYSTEM or NETWORK SERVICE accounts, to get Kerberos authentication, ServerSPN can be in the MachineName$@Domain format.)
A user-specified SPN.
KDCKey is an alphanumeric string that conforms to the rules for a KDC key.
For syntax-specific information, see the following topics:
For information about sample applications that demonstrate this feature, see SQL Server Data Programmability Samples.