> lower volume and few servers caused uneven load, and we went back to least connections for those clusters. The more traffic you’re handling, the more the cen

> lower volume and few servers caused uneven load, and we went back to least connections for those clusters. The more traffic you’re handling, the more the central limit theorem applies as you are summing the behavior of lots of random events drawn from various random distributions, and the more the system behavior regresses to the mean. I think of it as analogous to how when an engine is idling it tends to pop and judder, but as you rev it up the smoother it sounds until at sustained high revs it can make almost a pure note. All the variations are smoothed out and the system runs much more evenly and predictably, but when it drops to idle speed the subtle variation in the fuel air mix and the pressure in the exhaust and inlet every stroke makes it chug and sputter chaotically.