In many SQL Server workloads, especially OLTP, the database’s transaction log can be a bottleneck that adds to the time it takes a transaction to complete. Most people assume that the I/O subsystem is the real bottleneck, with it not being able to keep up with the amount of transaction log being generated by the workload.
I recently encountered high TRANSACTION_MUTEX accumulated wait time on a client system. I couldn’t recall a case where I saw this wait type as near the top of the “high waits” list and I was curious about what factors could push this type of overall wait time.
The Books Online definition of TRANSACTION_MUTEX is that it "occurs during synchronization of access to a transaction by multiple batches." Not many areas within the SQL Server engine expose this type of functionality, so my investigation was narrowed down to the following technologies: