Cheetah: A High-Speed Programmable Load-Balancer Framework With Guaranteed Per-Connection-Consistency

Large service providers use load balancers to dispatch millions of incoming connections per second towards thousands servers. There are two basic yet critical requirements for a balancer: uniform distribution the across servers, which requires support advanced balancing mechanisms, and per-connection-consistency (PCC), i.e, ability map packets belonging same connection server even in presence changes number active servers balancers. Yet, simultaneously meeting these has been an elusive goal. Today’s minimize PCC violations at price non-uniform distribution. This paper presents Cheetah, balancer that supports mechanisms and while being scalable, memory efficient, fast processing packets, offers comparable resilience clogging attacks as with today’s The Cheetah LB design guarantees any realizable selection mechanism can be deployed both stateless stateful manners, depending on operational needs. We implemented software Tofino-based hardware switch. Our evaluation shows version PCC, negligible packet overheads, reduce flow completion time by factor $2-3 \times $ .