Why Cloudflare chose leaderless QuePaxa for Meerkat’s global consensus service

Cloudflare says its experimental Meerkat service uses QuePaxa, a newer consensus algorithm, because Meerkat must operate across continents where network latency and connectivity can vary. Unlike leader-based approaches such as Raft, QuePaxa is designed to make progress without relying on a single leader; the article contrasts this with a 2020 network-switch failure that made a consensus-backed dashboard 80 times slower for six hours and unavailable for writes, without data loss.

For globally distributed control-plane services, QuePaxa’s leaderless design could reduce stalls caused by a slow or unreachable leader, but Cloudflare describes Meerkat as experimental and the article provides no performance measurements or operational comparison with Raft. Teams considering it should weigh that potential resilience against the uncertainty of adopting a newer algorithm.