ReferenceGlossary
Fallacies of distributed computing
Eight false assumptions that programmers new to distributed systems often make, collected by L. Peter Deutsch and others at Sun Microsystems: the network is reliable, latency is zero, bandwidth is infinite, the network is secure, topology does not change, there is one administrator, transport cost is zero, and the network is homogeneous.
Network
In Quorum#
Several Incidents and costs break these assumptions. In the Network partition Incident, calls between regions time out until the partition heals or you fence the old leader. A call to another zone adds 1 ms and costs egress both ways, and a call to another region pays the round trip. Until Timeouts and retries, every Link waits 30 s for an answer.