Four isolation levels: Read Uncommitted (weakest), Read Committed (default), Repeatable Read (stronger), Serializable (strongest)

Three concurrency anomalies: Dirty reads (read uncommitted data), non-repeatable reads (same query different results), phantom reads (new rows appear)

Isolation level trade-offs: Consistency vs performance, stronger isolation = more locks = slower queries

Setting isolation levels: SET TRANSACTION ISOLATION LEVEL syntax for PostgreSQL, MySQL, SQL Server, Oracle

Read Uncommitted: Fastest but allows dirty reads, rarely used in production (approximate counts only)

Read Committed: Default in most databases, prevents dirty reads, good balance between consistency and performance

Repeatable Read: Prevents non-repeatable reads, same query returns same results within transaction, used for reports

Serializable: Strongest isolation, transactions execute as if sequential, prevents phantom reads, slowest performance

Four isolation levels: Read Uncommitted (weakest), Read Committed (default), Repeatable Read (stronger), Serializable (strongest)

Three concurrency anomalies: Dirty reads (read uncommitted data), non-repeatable reads (same query different results), phantom reads (new rows appear)

Isolation level trade-offs: Consistency vs performance, stronger isolation = more locks = slower queries

Setting isolation levels: SET TRANSACTION ISOLATION LEVEL syntax for PostgreSQL, MySQL, SQL Server, Oracle

Read Uncommitted: Fastest but allows dirty reads, rarely used in production (approximate counts only)

Read Committed: Default in most databases, prevents dirty reads, good balance between consistency and performance

Repeatable Read: Prevents non-repeatable reads, same query returns same results within transaction, used for reports

Serializable: Strongest isolation, transactions execute as if sequential, prevents phantom reads, slowest performance