Hi, we have a use case to build a Cassandra cluster across 10 to 20 different data centers (many are smaller satellite DC with racks to support the local Cassandra ring, and with fast network connections). The eventual-consistent model fits well, as we tolerate minutes or even an hour before all data centers to reach data consistency.
I have done 3 DCs C* cluster in the past. I also saw published case of 4 DCs across continents (see link https://cloudmanagement.navisite.com/use-case-cassandra-for-distributed-nosql-database-across-multiple-data-centers/ <https://cloudmanagement.navisite.com/use-case-cassandra-for-distributed-nosql-database-across-multiple-data-centers/>). But 3 or 4 DCs are a smaller number compared to 20 DCs. Can you share the experience with Cassandra Cluster across tens of DCs? How 's the replication latency across the network, and have you seen scalability limitations? Thanks, Jim