A data transfer protocol utilizes a set of actions for affecting different objects that are stored in a data store. The data transfer protocol supports a set of application programming interfaces (APIs) that are applicable to the different objects. With the APIs, the data transfer protocol may synchronize a centralized data store with independent data stores. The data transfer protocol may support data that is hierarchical while maintaining referential integrity. Node objects may vary from nightly bulk uploads to interactive users uploading or querying smaller portions of the database. The data transfer protocol provides the semantics to carry out these operations with a fixed number of APIs for any arbitrary database. The data uploads may be sparse depending on what the remote data store holds and need not be in serial order.