ARCHITECTURE Signal 152
Cloudflare Introduces Meerkat for Strongly Consistent Global Coordination
↗
For engineers operating multi-region systems, leader-based consensus like Raft creates availability gaps when a leader crashes or becomes unreachable, blocking writes until timeout-based failover completes. Meerkat's asynchronous approach eliminates that single point of failure, though the trade-off is one to three round trips per write decision, raising questions about whether performance remains competitive under normal conditions.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Meerkat is based on QuePaxa, an asynchronous consensus algorithm that does not rely on timeouts, making it the first production deployment of such an algorithm at global scale.
Unlike Raft, QuePaxa allows all replicas to accept writes without a leader, avoiding the availability loss that occurs when a leader becomes unreachable across a wide-area network.
Meerkat is not designed for general-purpose databases and requires one to three round trips between a proposer and a majority of replicas to decide on a proposal, which practitioners flagged as a potential latency concern.
THE CLUSTER