Implementing Concurrency Control

1. Designing Thread-Safe Services

Pattern	Detail
Stateless	Easiest; default for Spring beans
Immutable state	Safe to share
Confined state	ThreadLocal
Synchronized	Guarded mutation
Concurrent collections	`ConcurrentHashMap`

2. Implementing Synchronization Mechanisms

Tool	Use
`synchronized`	Simple monitor lock
`ReentrantLock`	Try-lock, timeouts, fairness
`ReadWriteLock`	Many readers / 1 writer
`StampedLock`	Optimistic reads
`volatile`	Visibility, no atomicity
`AtomicInteger/Reference`	Lock-free counters

3. Using Concurrent Data Structures

Structure	Use
`ConcurrentHashMap`	General concurrent map
`CopyOnWriteArrayList`	Read-heavy, write-rare
`BlockingQueue`	Producer/consumer
`ConcurrentLinkedQueue`	Lock-free FIFO
`ConcurrentSkipListMap`	Sorted concurrent map

4. Implementing Async Processing

Example: Spring @Async

@Configuration @EnableAsync
public class AsyncConfig {}

@Service
public class EmailService {
    @Async("emailExecutor")
    public CompletableFuture<Void> sendWelcome(User u) {
        client.send(u);
        return CompletableFuture.completedFuture(null);
    }
}

5. Implementing Idempotency Keys

Concern	Detail
Concurrent retries	Use unique key + insert
DB constraint	UNIQUE on idempotency_key
Catch dup	Return existing result

6. Handling Race Conditions

Type	Fix
Check-then-act	Atomic compare-and-set
Read-modify-write	Optimistic lock or single SQL
Lost update	`@Version` / SELECT FOR UPDATE
Double execution	Idempotency / distributed lock

7. Designing Deadlock Prevention

Strategy	Detail
Lock ordering	Always acquire in same order (e.g., by ID)
Timeouts	`tryLock(timeout)`
Lock-free	Use atomics, immutables
Coarser locks	Sometimes simpler than fine-grained

8. Implementing Thread Pools

Example: Bounded executor

ExecutorService exec = new ThreadPoolExecutor(
    /* core */ 8, /* max */ 32,
    60, TimeUnit.SECONDS,
    new ArrayBlockingQueue<>(1000),
    new ThreadFactoryBuilder().setNameFormat("worker-%d").build(),
    new ThreadPoolExecutor.CallerRunsPolicy());  // backpressure

Warning: Avoid Executors.newCachedThreadPool() in production — unbounded threads can OOM.

9. Using Reactive Programming

Why	Detail
Non-blocking	Fewer threads needed
Backpressure	Producer respects consumer rate
Composable	Operators chain
Cost	Steeper learning curve, harder debugging

10. Implementing Read-Write Locks

Example: ReentrantReadWriteLock

private final ReadWriteLock rw = new ReentrantReadWriteLock();
public V get(K k) { rw.readLock().lock(); try { return map.get(k); } finally { rw.readLock().unlock(); } }
public void put(K k, V v) { rw.writeLock().lock(); try { map.put(k, v); } finally { rw.writeLock().unlock(); } }

11. Implementing Semaphores

Example: Concurrency limit

private final Semaphore inflight = new Semaphore(50);
public Result call() throws InterruptedException {
    inflight.acquire();
    try { return downstream.invoke(); }
    finally { inflight.release(); }
}

12. Implementing Lock-Free Algorithms

Tool	Detail
`AtomicReference`	CAS-based updates
`LongAdder`	High-throughput counter
`ConcurrentHashMap.compute*`	Atomic compound operations
Disruptor	Lock-free ring buffer

Note: Java 21 virtual threads NEW let you write blocking code at scale — often a simpler alternative to reactive.