I'll tell about new best practices of large Node.js apps, which allow to increase throughput by 10X times.
The growth of functionality in Enterprise applications often leads to a decline in throughput. We start with the dispersion of the Main Thread and spread it into a spectrum of operations. We learn which of these operations we can offload from the Main Thread to threads/workers/processes, and how to do that. We select successful ways of parallelization for each type of operation, consider the benefits and limitations. Compare the throughput after the Main Thread decomposition.