[concurrent programming] synchronization container, concurrent container, blocking queue, double ended queue and work secret

1、 Synchronization container

The main representatives are Vector and Hashtable, as well as Collections.synchronizedXxx etc. .

The granularity of the lock is the current object as a whole .

Iterators fail in time , That is to say, it is found to be modified in the process of iteration , Will throw ConcurrentModificationException.

2、 Concurrent container

The main representatives are ConcurrentHashMap、CopyOnWriteArrayList、ConcurrentSkipListMap、ConcurrentSkipListSet

The granularity of locks is decentralized 、 Fine grained , That is, reading and writing use different locks .

Iterators have weak consistency , That is, it can tolerate concurrent modification , Don't throw ConcurrentModificationException.

3、 Blocking queues

The main representatives are LinkedBlockingQueue、ArrayBlockingQueue、PriorityBlockingQueue(Comparable,Comparator)、SynchronousQueue.

Provides blocking put and take Method , And support timed offer and poll Method .

Applicable to producers 、 Consumer model ( Thread pools and work queues -Executor)

It is also a synchronization container

4、 Double ended queue and work secret

The main representatives are ArrayDeque and LinkedBlockingDeque.

significance ： Just as blocking queues apply to producer consumer patterns , Double ended queue is also applicable to another mode , That is, work secret . In the producer - In consumer design , All consumers share a work queue , And in working secret , Each consumer has its own double ended queue .

If a consumer has completed all the work in their double ended queue , Then it can secretly get work from the end of the double ended queue of other consumers . Better scalability , This is because worker threads do not compete on a single shared task queue .

Most of the time , They all just access their own double ended queues , Thus greatly reducing the competition . When a worker thread needs to access another queue , It gets work from the end of the queue instead of the head , Therefore, the competition on the queue is further reduced .