Developed for fine-grained and high-concurrency thread control, mutexes are implemented at a very low system call level, whereas latches rely more on Oracle kernel code. Each mutex call starts with the mutex abbreviation mtx, and the call names should come as no surprise. Here is a partial list of mutex calls: initialization (mtx_init), destruction (mtx_destroy), spinning (mtx_lock_spin, mtx_unlock_spin), and sleeping (mtx_sleep).

Because mutexes are based on standard system calls, they form a base functionality on which Oracle can build to meet its specific objectives. For example, suppose Oracle wants greater control then the standard mutex spin call provides. Instead of using only the mutex spin request call, Oracle can add code to meet its objectives. This gives Oracle a tremendous amount of flexibility, and also allows Oracle to more easily integrate new facilities, like mutexes, into its existing code base.

There are a few very important and remarkable reasons why Oracle chose to start using mutexes. Besides a smaller memory footprint6 than latches (16 bytes compared to 112 bytes), mutexes give Oracle kernel developers more control over structure creation, reduce false contention, reduce the likelihood of the control structure being the bottleneck, and provide faster pinning times. Let's take a closer look at why Oracle was motivated to begin using mutexes.