SSPI security exit

IBM MQ for Windows supplies a security exit for both the IBM MQ MQI client and the IBM MQ server. This is a channel-exit program that provides authentication for IBM MQ channels by using the Security Services Programming Interface (SSPI). The SSPI provides the integrated security facilities of Windows systems.

The security packages are loaded from either security.dll or secur32.dll. These DLLs are supplied with your operating system.

One-way authentication is provided using NTLM authentication services. Two-way authentication is provided using Kerberos authentication services.

The security exit program is supplied in source and object format. We can use the object code as it is, or we can use the source code as a starting point to create your own user-exit programs.

See also Use the SSPI security exit on Windows.

Introduction to security exits

A security exit forms a secure connection between two security exit programs, where one program is for the sending message channel agent (MCA), and one is for the receiving MCA.

The program that initiates the secure connection, that is, the first program to get control after the MCA session is established, is known as the context initiator. The partner program is known as the context acceptor.

The following table shows some of the channel types that are context initiators and their associated context acceptors.

Context Initiator	Context Acceptor
MQCHT_CLNTCONN	MQCHT_SVRCONN
MQCHT_RECEIVER	MQCHT_SENDER
MQCHT_CLUSRCVR	MQCHT_CLUSSDR

The security exit program has two entry points:

• SCY_NTLM

  This uses NTLM authentication services, which provide one-way authentication. NTLM allows servers to verify the identities of their clients. It does not allow clients to verify a server's identity, or one server to verify the identity of another. NTLM authentication was designed for a network environment in which servers are assumed to be genuine.

• SCY_KERBEROS

  This uses Kerberos mutual authentication services. The Kerberos protocol does not assume that servers in a network environment are genuine. Parties at both ends of a network connection can verify the identity of the other party. That is, servers can verify the identity of clients and other servers, and clients can verify the identity of a server.

What the security exit does

This topic describes what the SSPI channel-exit programs do.

The supplied channel-exit programs provide either one-way or two-way (mutual) authentication of a partner system when a session is being established. For a particular channel, each exit program has an associated principal (similar to a user ID, see IBM MQ access control and Windows principals ). A connection between two exit programs is an association between the two principals.

After the underlying session is established, a secure connection between two security exit programs (one for the sending MCA and one for the receiving MCA), is established. The sequence of operations is as follows:

1. Each program is associated with a particular principal, for example as a result of an explicit login operation.
2. The context initiator requests a secure connection with the partner from the security package (for Kerberos, the named partner) and receives a token (called token1). The token is sent, using the underlying session that is already established, to the partner program.
3. The partner program (the context acceptor) passes token1 to the security package, which verifies that the context initiator is authentic. For NTLM, the connection is now established.
4. For the Kerberos-supplied security exit (that is, for mutual authentication), the security package also generates a second token (called token2), which the context acceptor returns to the context initiator by using the underlying session.
5. The context initiator uses token2 to verify that the context acceptor is authentic.
6. At this stage, if both applications are satisfied with the authenticity of the partner's token, the secure (authenticated) connection is established.

IBM MQ access control and Windows principals

The access control that IBM MQ provides is based on the user and group. The authentication that Windows provides is based on principals, such as user and servicePrincipalName (SPN). In the case of servicePrincipalName, there might be many of these associated with a single user.

The SSPI security exit uses the relevant Windows principals for authentication. If Windows authentication is successful, the exit passes the user ID that is associated with the Windows principal to IBM MQ for access control.

The Windows principals that are relevant for authentication vary, depending on the type of authentication used.

• For NTLM authentication, the Windows principal for Context Initiator is the user ID associated with the process that is running. Because this authentication is one way, the principal associated with the Context Acceptor is irrelevant.
• For Kerberos authentication, on CLNTCONN channels, the Windows principal is the user ID associated with the process that is running. Otherwise, the Windows principal is the servicePrincipalName that is formed by adding the following prefix to the QueueManagerName.

  ibmMQSeries/
  

Parent topic: Building your procedural application on Windows