I am not sure if READ_COMMITTED will solve your issue fully.
I would lean towards using optmistic locking instead.
Unfortunately, optimistic locking support does not exist in 2.4. It has been fixed in 2.6 in addition to other locking issues in a clustered setup.
We are still striving to make it even better.
If possible, try upgrading to our latest portal release which is 2.6.3