Problem:
How and when use Java Fixed Thread Pool? In this post we’ll show the ExecutorService in action and describe its characteristics.
Solution:
java.util.concurrent.Executors class has a number of static factory methods to create predefined ExecutorServices. There are two methods to create executors with Fixed Thread Pool:
- ExecutorService newFixedThreadPool(int nThreads)
Creates a thread pool with nThreads and executes at most as many tasks. - ExecutorService newFixedThreadPool(int nThreads, ThreadFactory threadFactory)
Creates a thread pool with nThreads and executes at most as many tasks. When needed new threads are created using provided ThreadFactory.
In the following example we’ll use the first method to create a pool that executes at most two tasks at the same time. If you run the example you will see that they report count in pairs:
package com.farenda.java.util.concurrent;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import static java.util.concurrent.TimeUnit.MILLISECONDS;
public class FixedThreadPoolExecutorExample {
private static class Counter implements Runnable {
private static int nth = 0;
private final int id = ++nth;
private final int number;
public Counter(int number) {
this.number = number;
}
@Override
public void run() {
System.out.println("Starting counter: " + id);
for (int i = 0; i < number; ++i) {
try {
MILLISECONDS.sleep(300);
} catch (InterruptedException e) {
// ignore
}
System.out.printf("counter %d, value: %d%n", id, i);
}
System.out.println("Finishing counter: " + id);
}
}
public static void main(String[] args) {
int nThreads = 2;
System.out.printf("Starting Fixed Thread Pool of %d threads%n", nThreads);
ExecutorService executor = Executors.newFixedThreadPool(nThreads);
for (int i = 1; i <= 4; ++i) {
executor.execute(new Counter(3));
}
executor.shutdown();
}
}
[/sourcecode]
<p> And here's the result of running the above code: </p>
Starting Fixed Thread Pool of 2 threads
Starting counter: 1
Starting counter: 2
counter 1, value: 0
counter 2, value: 0
counter 1, value: 1
counter 2, value: 1
counter 1, value: 2
Finishing counter: 1
Starting counter: 3
counter 2, value: 2
Finishing counter: 2
Starting counter: 4
counter 3, value: 0
counter 4, value: 0
counter 3, value: 1
counter 4, value: 1
counter 3, value: 2
Finishing counter: 3
counter 4, value: 2
Finishing counter: 4
As you can see the threads (and hence tasks) are executed in pairs, which exactly matches the size of our Fixed Thread Pool.
If a thread dies then the pool will create a new one to match specified size.
Fixed Thread Pool creates the threads when it starts and doesn’t waist time later when tasks are submitted. Also, due to its fixed size, there’s no danger going out of number of threads limit, which could in other executors when many tasks would be submitted in short period of time.
