I'm confused by mono table (which implies a single table) and how mono doesn't mean single point of failure. They talk about a cluster, and adding machines, so

I'm confused by mono table (which implies a single table) and how mono doesn't mean single point of failure. They talk about a cluster, and adding machines, so I guess mono table lives on multiple machines... but how? Is it replicated to every machine? (EG: every node in the cluster contains a complete set of the data?) IF so, this would mean that as you scale, the updates of writes to replicate the changes across the cluster could possibly become quite a problem, in terms of bandwidth and latency. I can imagine each node spending most of its time handling updates from other nodes. Also, unless you're locking the table on every node simultaneously (the thought of dealing with that gives me headaches) you're going to have consistency issues. I think their goals are excellent-- zero admin, no SPOF are really important features to me. If they've got a new architecture, then it could be a really important and innovative project. Would love to see an architectural description.