Pessimistic locking Vs Optimitic locking In ERP

Question

Hi all
" Pessimistic locking Vs Optimistic locking To advice MY ERP "
Im SASI , Now I have a problem for using locking system in update problem
I am using Oracle 10g and Java 5.0(Netbeans)
My PL to advice for pessimistic locking system ,
Now I want to what is the difference between & How to help my ERP
I am reading lot of information they are all confuse to me.
1) Select .. for update nowait --> this statement will lock that record upto the user commit,
in my problem I am wait for the same record that fellow when will release I'm updated.
2) Do the modified (I have column upto 25 ) how to change to another transaction in some other table
ex orderdetails -order cancel
stock detail -stock maintain
orderdetails for update ........................... long time at the same time waiting for stock detail updatation .
ALL the guys How to handle the above top and I want solution above problem
very urgent .....
Thank
SASIKUMARA

Answer 1

Pessimistic Locking:
You lock any and all rows before you update them. In Oracle that is done with a SELECT FOR UPDATE.
Problem:
Stateless clients (like web clients/app tier connections) cannot use pessimistic locking as this locking is specific to a single Oracle session. These stateless clients can use/re-use/create any Oracle session. Thus should they attempt to lock rows in one session, that state information will be missing when they attempt to update the rows from another session.
Work-around - Optmistic Locking:
To work around this problem with stateless clients, optimistic locking is used. The client does not lock the rows it wants to update. It optimistically assume that the rows will be unchanged at the time it wants to update them.
When it does update the rows, it needs to check the existing copy of the row in the database, with the old copy of the row it retrieved from the database. If these two copies matches, then it can proceed to write the changed copy of the row.
This is usually done via an UPDATE SQL like this:
UPDATE emp
SET name = :newname, surname = :newsurname, ...
WHERE emp_id = :emp_id
AND name = :oldname
AND surname = :oldsurname
AND ... (for all other columns) As you can see, should any column in the row be changed from the last time that the client viewed the row, the UPDATE will fail to update the changed row.
Not implementing optimistic locking for stateless clients leads to issues like lost updates. Two clients update the same row - the 2nd client overwrites the changes of the 1st client and thus the 1st client's changes becomes lost.