Concurrent Update Problems

When many people are concurrently updating data, four problems can arise:

• Lost updates. One user's update overlays another user's update, causing it to be lost.

• Dirty data. One user sees another user's uncommitted changes.

• Non-repeatable reads. When a record is read for the second time during a transaction, its values have changed.

• Phantoms. Records appear or disappear during a transaction.