Kaleido is a novel data system for multi-device synchronization. According to the theories of conflict-free replicated data types, Kaleido enables clients to synchronize replicated data types with operations, and gurantees high availability and eventual consistency. Implemented as a back-end as a service, Kaleido server is scalable so as to deal with a huge amount of requests from massive clients. In bringing academic achievements into the real world data system, several issues and adaptations arise, and this paper introduces them.