The system and method described herein provide an order fulfillment database system. The order fulfillment database system may provide a high throughput speed, high availability, redundancy, and scalability. The order fulfillment database system supports very fast transaction times for atomicity, consistency, isolation, and durability (ACID) transactions. Order services instances at a primary cluster provide in-memory databases that service segments of the persistent database and are associated with different order fulfillment entities. A routing manager routes database messages to the appropriate order services instance. A secondary cluster of order services instances provides redundancy. A retry procedure preserves the integrity of the database transactions based on a current operational state of the primary order services instances.