Handling Concurrency and Preventing Duplicate Ticket Booking in Train Ticket Systems

In a train ticket ordering scenario, two users (A and B) simultaneously view an available lower berth; user B books it first, and when user A later attempts to book the same berth, the system reports it as already reserved, highlighting the need for proper concurrency control.

Solution 1 – Simple Flag Field: Add a FLAG column (0 = free, 1 = reserved) to the seat table and check it before booking. Under high load, race conditions occur: both users read FLAG=0, both set it to 1, resulting in duplicate reservations. The timeline is illustrated as: A  FLAG=0  at T1 (query) B  FLAG=0  at T2 (query) B  FLAG=1  at T3 (update) A  FLAG=1  at T4 (update)

Solution 2 – Pessimistic Lock (FOR UPDATE SKIP LOCKED): Use a query such as SELECT * FROM table WHERE … FOR UPDATE SKIP LOCKED to lock rows during selection; if a row is already locked, the query skips it. While this prevents duplicate bookings, it degrades user experience because available seats may be hidden.

Solution 3 – Synchronized Block: Apply synchronized in application code to serialize access. This works only in a single‑node deployment; in a clustered environment a global lock mechanism is required.

Solution 4 – Hybrid Optimistic‑Pessimistic Lock: First read the seat information without locking (optimistic read). When a user proceeds to book, acquire a pessimistic lock on that specific record with SELECT * FROM table WHERE … FOR UPDATE . The second user attempting to lock the same record will fail, thus preventing duplicate booking.

The article concludes that combining optimistic reads with targeted pessimistic locks offers a practical balance between concurrency and user experience in high‑traffic ticketing systems.