Symmetric Multiprocessing with MySQL - MySQL

Threading the Needle

To better understand threads in general, we must look at the relationship between threads and processes. A process contains both an executing program and a collection of resources, such as the file map and address space. All threads associated with a given task share the task’s resources. Thus, a thread is essentially a program counter, a stack, and a set of registers; all the other data structures belong to the task. A process effectively starts out as a task with a single thread.

A thread is a miniprocess that has its own stack and that executes a given piece of code. A thread normally shares its memory with other threads, unlike true processes, which will usually have a different memory area for each one. A thread group is a set of threads all executing inside the same process. They all share the same memory and thus can access the same global variables, the same heap memory, the same set of file descriptors, and so on. All the threads in a thread group execute either by using time slices when running on a single processor or in parallel if multiple processors are available.

The advantage of using a thread group over using a process group is that context switching between threads is much faster than context switching between processes; in other words, the system switches from running one thread to running another thread much faster than it can switch from one process to another. Also, communication between two threads is usually faster and easier to implement than communication between two processes, since the threads already share common address space in which to share variables.

The POSIX thread libraries are a standards-based thread API for C and C++. They are most effectively used on multiprocessor systems, where the process flow can be scheduled to run on another processor, thus increasing speed through parallel or distributed processing. Threads require less overhead than forking, or spawning a new process, because the system will not initialize a new virtual memory space and environment for the process. While POSIX is most beneficial on a multiprocessor system, gains are also found on single processor systems, which exploit latency in input/output (I/O) and other system functions that can halt process execution.