allocationSize

allocationSize is a configuration parameter used by certain persistence frameworks to optimize the generation of identifier values that rely on database sequences or table-based generators. It determines how many identifiers are reserved and cached by the application when generating new entities, reducing the number of times the database must be consulted.

In Java Persistence API (JPA) contexts, allocationSize is commonly used with sequence- or table-based generators such as @SequenceGenerator and @TableGenerator. For example, a generator defined with allocationSize=50 will reserve a block of 50 IDs from the underlying sequence or table. The persistence provider then assigns these IDs to new entities in memory until the block is exhausted, at which point it requests another block from the database.

How it works in practice: the database sequence (or table) maintains a numeric value that is advanced when a new block is requested. The application uses the block locally, which minimizes round-trips to the database during bulk inserts. If the application terminates unexpectedly or if the block is not fully consumed, there can be gaps in the observed sequence values, although the IDs remain unique and valid.

Choosing a value involves trade-offs between performance and potential value waste. A larger allocationSize reduces database access but can increase the number of unused IDs if the application restarts or fails before consuming the block. A smaller value increases database traffic but minimizes waste. In distributed deployments, keep the allocationSize consistent with the database sequence increment to avoid conflicts.

Alternatives include identity or UUID generation strategies, where allocationSize does not apply.