How deeply does Java Persistence with Spring Data and Hibernate epub cover repeatable read isolation level?

Compares prevented anomalies, costs, and related JPA locking behavior, without specifying a database's internal isolation algorithm.

Repeatable read isolation level

As explained in Java Persistence with Spring Data and Hibernate epub

Repeatable read is a transaction isolation level under which repeated reads of the same data return a consistent value within a transaction, preventing unrepeatable reads while generally allowing phantom reads.

Repeatable read is a transaction isolation level for keeping a row's value stable across repeated reads within one transaction. It prevents an unrepeatable read: reading a value, then reading a different value after another transaction updates and commits it. It also prevents dirty reads and lost updates in the ANSI isolation-level comparison, but it does not prevent phantom reads. A repeated query can therefore return additional or fewer rows after another transaction inserts or deletes data. section 11.2.1

Table 11.1 ANSI isolation levels and the problems that they address (view table figure)

Isolation level	Phantom read	Unrepeatable read	Dirty read	Lost update
`READ_UNCOMMITTED`	–	–	–	+
`READ_COMMITTED`	–	–	+	+
`REPEATABLE_READ`	–	+	+	+
`SERIALIZABLE`	+	+	+	+

The isolation-level table compares which concurrency problems are prevented at each level: repeatable read rules out unrepeatable reads, while serializable also rules out phantom reads. section 11.2.1

Choose the level with the transaction's required guarantees in mind. Higher isolation has performance and scalability costs. A JPA pessimistic lock is a separate way to protect selected data: PESSIMISTIC_READ guarantees repeatable reads, while PESSIMISTIC_WRITE additionally requires serialized access without phantom reads. Hibernate's locking SQL depends on the database dialect, and the lock lasts only for the database transaction. section 11.2.1 section 11.2.3

Lock rows used in a calculation

A transaction totals item prices across categories by querying each category with setLockMode(LockModeType.PESSIMISTIC_READ) and a lock-timeout hint. Hibernate adds database-specific locking SQL: PostgreSQL can use FOR SHARE, whereas H2 uses FOR UPDATE for its pessimistic modes. If a required lock cannot be obtained, the operation throws an exception. section 11.2.3

Do not mistake a repeatable managed-entity result for transaction-level repeatable read. Without an active transaction, Hibernate can return an already-loaded entity from the persistence context, preserving its in-memory value, while a scalar query reads the database value in auto-commit mode. Database isolation also matters when refreshing entities: MySQL's stated default is REPEATABLE_READ, so an example that must see another transaction's committed changes configures the JDBC connection for READ_COMMITTED. section 11.3.1 section 13.1.3

How Java Persistence with Spring Data and Hibernate epub covers it

Cover of Java Persistence with Spring Data and Hibernate epub

liveBook manning.com

Read Java Persistence with Spring Data and Hibernate epub on liveBook Get the book at manning.com

Also explained in

liveBook manning.com

100 SQL Server Mistakes and How to Avoid ThemRepeatable Read is a transaction isolation level that prevents dirty and nonrepeatable reads but can allow phantom reads.

Compare how 2 books explain repeatable read isolation level →

Related concepts

isolationkind of · Isolation is the separation of components, processes, resources, or contexts to limit interaction, interference, or the propagation of effects between them.
nonrepeatable readaddresses · A nonrepeatable read is a concurrency anomaly in which repeated reads of the same data within a transaction return different values after another transaction modifies it.
MySQLconfigured by · MySQL is an open-source relational database management system that stores, manages, and queries structured data using SQL.
PESSIMISTIC_WRITEproduced by · PESSIMISTIC_WRITE is a JPA lock mode that acquires an exclusive lock, guarantees repeatable reads, and serializes conflicting data access.
database transactionconfigures · A database transaction is a logical unit of database operations whose changes are committed together or rolled back as a whole.
read committed isolation levelcontrasts with · A database transaction isolation level that prevents reads of uncommitted changes while allowing later reads in a transaction to observe newer committed data.

This explainer was generated by Manning's knowledge graph from the text of Java Persistence with Spring Data and Hibernate epub. The section links point to the parts of the book it draws on.