Let's say you have built an email database. Marketing, ML team pumps in tons of writes and you use it while sending out emails. You have a account db that hold

Let's say you have built an email database. Marketing, ML team pumps in tons of writes and you use it while sending out emails. You have a account db that holds on the user information. What is AWESOME, you can create a topology like: [accountDB,replicas] <- [emailDB,replicas]. Where emailDB only replicates the user table from accountDB using replication filters [1]. This way anyone writing a service can just do normal joins [select * from email,user where user.id=email.user_id] etc ON the emailDB. https://dev.mysql.com/doc/refman/5.7/en/change-replication-filter.html https://dev.mysql.com/doc/refman/5.7/en/change-replication-f...