Explain the differences between phantom reads and non-repeatable reads in the context of transactions and data isolation.

In the context of transactions and data isolation:

• Non-repeatable Read occurs when, within a single transaction, the same data query returns different results because another transaction has modified and committed those data between the two reads.

• Phantom Read is a situation where, during a transaction, a repeated query with a condition (e.g., selecting rows based on a criterion) reveals new rows that were not visible during the first read because another transaction inserted or deleted rows matching the condition.

Main difference:

• Non-repeatable read is related to changes in the values of already read rows.
• Phantom read is related to the appearance or disappearance of new rows matching the selection condition.

Example:

1. Transaction A reads all customer orders with an amount > 1000.
2. Transaction B adds a new order with an amount of 1500 and commits.
3. Transaction A repeats the same query and sees the new "phantom" order.

To prevent these phenomena, transaction isolation levels are used, for example, Repeatable Read prevents non-repeatable reads, and Serializable prevents phantom reads.