Only one user thread can access the kernel at a time

The server creates a thread for each request

The thread is discarded once it has completed its work

Helps developing modern applications with hundreds of threads

Programmers identify tasks that can run in parallel

The programmer writes a task as a function

Too many threads that work concurrently can exhaust system resources

In explicit threading The number of threads that can be created has no upper bound.

in implicit threading Programmers identify threads

Run-time libraries (not programmers) create and manage threads

works well for tasks that are synchronous (cooperating).

works well for asynchronous (independent) tasks

in implicit threading Each task is mapped to a separate thread using the many-to-many model

At startup, the server creates a set of threads in a pool

Implicit fork-join is an asynchronous version of the thread pool

Create one or more child threads. then Wait for the created threads (children) to terminate. finally Join the threads, at which point it can retrieve and combine their results

When the server receives a request, it submits it to the pool

If the pool has a thread, the …. is served immediately

in Thread pool If the pool is empty, the task is queued until a thread becomes free

in Thread pool Once a thread completes serving the request, it returns to the pool.

…. can adjust the number of threads in the pool.