Performance optimization techniques in time series databases: Limiting concurrency

The canonical way to limit the amount of concurrency in your application is to spawn a limited number of worker threads and share a channel between them for dispatching work. Since the channel has a limited buffer size, only a limited number of workers can run at once. Your application can spawn a worker goroutine whenever a unit of work comes in and concurrencyLimitCh does the hard work of guaranteeing that no more than some maximum number of workers are running at once.