I have read through the user guide and transaction, locks, and asynchronous API, and I have some  questions about these topics.

Actually all my questions are to ensure we correctly handle a transaction that versioning is really critical, and get much concurrency as possible. It can be simplified as following code.

tx.begin();    //(a)

cache.getAdvancedCache().lock(k); //(b)

Integer v=cache.get(k); //(c)

cache.putAsync(k, v + increasement); //(d)

tx.commit();  //(e)

So here are my questions.

1)  Regarding the API of AdvancedCache, we can load an object from cache through:

     cache.getAdvancedCache.withFlags(Flag.FORCE_WRITE_LOCK).get(key);

      what's the difference is this from

     cache.getAdvancedCache().lock(key);

     cache.get(key) ?

     Is the lock obtained by .withFlags(Flag.FORCE_WRITE_LOCK) also a cluster wide lock?

2) If using asynchronous api, what happens during a transaction prepare phase and commit phase? For example

     tx.begin()
     cache.putAsync(K,V5) 
     tx.commit()

    If eagerlocking is set to false, concluded from document about asynchronous API, I think a cluster wide lock will be acquired during the prepare phase. Will the replication complete before transaction comitted or after that?  It seems to be before transaction committed since the lock will be released after transaction committed, and we need this lock during replication. However, if the replication must be completed before transaction is committed, can we still get the benifit with asynchronous API?

3) During the prepare phase, if other transaction is processing commiting and has a lock on the same key in other node, will the XAResource wait for the lock to be granted util timeout? or will it fails the transaction immediately?

4) For my example, replace and put operation both works, which one is recommended for better performance.

Thanks in advance for your response.