Does paper-spanner somehow help with this issue mentioned in the cassandra paper?
p 1: Although strong consistency provides the application writer a convenient programming model, these systems are limited in scalability and availability [10] (#)
It seems you can't have strong consistency if you use optimistic replication techniques:
For systems prone to server and network failures, availability can be increased by using optimistic replication techniques, where changes are allowed to propagate to replicas in the background, and concurrent, disconnected work is tolerated (#)
So if this is true: Before a node is allowed to report that a transaction has committed, it must wait 7ms. (#) then do all nodes have to respond that the transaction completed before spanner proceeds?