@TanyaS58

Run the queue processor on dedicated BackgroundProcessing nodes and add more nodes if you need more throughput, making sure the QP is enabled on each node in Admin Studio. Increase max concurrent threads and tune batch size so CPUs stay busy without overloading the database or downstream APIs; raise these gradually while watching latency. In Queue-For-Processing use a good partition key to spread items evenly and avoid one hot key. Keep the payload tiny: queue only IDs or a few fields, then load data inside the processor with data pages; never queue big clipboard pages or attachments. Make the processing activity idempotent and short, set sensible timeouts, and avoid long loops or heavy joins. Cut down retries and route bad items quickly to the dead-letter queue with clear error handling so they don’t clog the pipe. If you still use legacy agents, add indexes to queue tables and purge old rows; for modern QPs watch stream lag and topic partitions. Keep an eye on Admin Studio metrics for throughput, failures, and delay, then adjust threads, partitions, and node count based on those numbers.