Systems and methods are described herein for creating, maintaining and applying a dataset that can be used to synchronize multiple instances of a data store. The dataset includes a sequentially-ordered sequence of change sets associated with the data store, wherein each change set includes zero or more entries that specify how the state of certain entities in the data store have changed since state changes represented by a previous change set in the sequence were applied to the data store. When a new change set is added to the dataset, any entries in any previously-added change sets dealing with the entities identified in the new change set are removed. Consequently, the dataset provides a concise and complete representation of all the state changes that have occurred to the data store since a known origin state.