Concurrency

Concurrency in the Java SDK is a worker-level setting: each worker runs its handlers on a thread pool, and the pool's size caps how many jobs that worker executes at once.

Worker worker = queue.worker()
        .handle(TRANSCODE, this::transcode)
        .concurrency(2) // at most 2 handlers run at once on this worker
        .start();

concurrency(0) (the default) uses a cached pool that grows with demand; any positive value fixes the thread count. Jobs beyond the cap stay PENDING and dispatch as running ones finish.

Autoscaling

Instead of a fixed size, let the pool track queue depth — it resizes between min and max threads, re-evaluated every couple of seconds:

Worker worker = queue.worker()
        .handle(TRANSCODE, this::transcode)
        .autoscale(AutoscaleOptions.of(2, 16)) // 2..16 threads, ~10 queued tasks per thread
        .start();

Scaling out

Total parallelism is the sum across workers — run more worker processes over the same storage to scale horizontally, and partition with queues(...) so a heavy task class gets its own capacity:

Worker videoWorker = queue.worker()
        .handle(TRANSCODE, this::transcode)
        .queues("video")
        .concurrency(4)
        .start();

Pair a bounded pool with pooled resources when handlers share an expensive client — the resource pool bounds the client instances while the thread pool bounds the executions.