Back when I grew up in Minnesota we played a game called "Duck, Duck, Gray Duck where one my friends would run around the circle of all us kids and tap us on the head - around and around. It's basically the same way a load balancing algorithm works to enable horizontal scalability of large systems. Except that the load balancer would not randomly tap one of us and say Gray Duck and then go sprinting around to take our spot. That's not actually how node eviction works in clustering. In this episode of PerfBytes we are joined by our old friend Alon Girmonsky from Blazemeter to discuss the common challenges with modern load balancers; both software and hardware types. Dan Bartow gives a rant about the past 7 years of bad load balancer performance. And James introduces a whole new type of pre-outage performance outage on News of the Damned. This episode of PerfBytes is formally sponsored by www.flood.io.