Suppose that we use the deferred update protocol for the example in Figure 23.6. Show how the log would be different in the case of deferred update by removing the unnecessary log entries; then describe the recovery process, using your modified log. Assume that only redo operations are applied, and specify which operations in the log are redone and which are ignored.

Question

What will be an ideal response?

Answer 1

In the case of deferred update, the write operations of uncommitted transactions are not
recorded in the database until the transactions commit. Hence, the write operations of T2
and T3 would not have been applied to the database and so T4 would have read the
previous (committed) values of items A and B, thus leading to a recoverable schedule.
By using the procedure RDU_M (deferred update with concurrent execution in a
multiuser environment), the following result is obtained:
The list of committed transactions T since the last checkpoint contains only transaction
T4. The list of active transactions T' contains transactions T2 and T3.
Only the WRITE operations of the committed transactions are to be redone. Hence, REDO
is applied to:
[write_item,T4,B,15]
[write_item,T4,A,20]
The transactions that are active and did not commit i.e., transactions T2 and T3 are
canceled and must be resubmitted. Their operations do not have to be undone since they
were never applied to the database.