> The author's suggestion of using synchronous communication with backpressure and sync failures is my last ditch approach Also back pressure isn’t difficult t

> The author's suggestion of using synchronous communication with backpressure and sync failures is my last ditch approach Also back pressure isn’t difficult to implement. Simply read the estimated size of the queue every N minutes and pass sending until it goes down to a more manageable level. Obvious downside is that it’s client side.