Abstract: Aspects of the subject matter described herein relate to auditing operations. In aspects, operations may be audited synchronously and/or asynchronously to one or more audit targets. When auditing synchronously, audit records may be written synchronously to an audit target. When auditing asynchronously, a buffer may be used to store audit records until the audit records are flushed to an audit target. If an error occurs in auditing, a policy may be evaluated to determine how to respond. One exemplary response includes failing an operation that triggered a subsequent audit record. Furthermore, if a buffer was unable to be copied to an audit target, the contents of the buffer may be preserved and one or more retries may be attempted to copy the buffer to the audit target.

Abstract: A trace header with an activity identifier is sent between a client and a database server as part of a request message stream. The activity identifier gets logged by client-side connection-related traces, by server-side traces, and by database engine traces. The activity identifier can be used to exactly correlate a thread on the client to threads on the server. Additionally, specific threads within a database engine may be correlated using the activity identifier to track the threads that processed a particular request. This enhances troubleshooting of the database system, particularly when there are many concurrent users and threads.

Abstract: A trace identifier is sent between a database client and a database server using a pre-login data package along with pre-login connection open requests through the TDS protocol. The trace identifier gets logged by client-side connection-related traces, by server-side traces, and by database engine traces. The trace identifier can be used to exactly correlate a physical connection on the client to a physical connection on the server. This enhances troubleshooting of the database system, particularly when there are many concurrent users and threads.

Abstract: Aspects of the subject matter described herein relate to auditing operations. In aspects, operations may be audited synchronously and/or asynchronously to one or more audit targets. When auditing synchronously, audit records may be written synchronously to an audit target. When auditing asynchronously, a buffer may be used to store audit records until the audit records are flushed to an audit target. If an error occurs in auditing, a policy may be evaluated to determine how to respond. One exemplary response includes failing an operation that triggered a subsequent audit record. Furthermore, if a buffer was unable to be copied to an audit target, the contents of the buffer may be preserved and one or more retries may be attempted to copy the buffer to the audit target.