> Finally, the third assumption we’ll make is that the CockroachDB cluster is operating under a steady-state that does not include chaos events.
> they shouldn’t represent the cluster’s typical behavior –– so, for the sake of this performance model, it’s safe to assume they are absent.
These events can be made somewhat rare within a datacenter. But they are actually very typical and very normal in communications between datacenters over the internet. Ignoring them is basically ignoring 90th percentile latency in such "geo distributed" configurations.
Which is fine, I always believed that Cockroach should focus on local clusters. But marketing it as "geo distributed" would just bring people disappointment, not necessarily a feeling you want potential customers to have.