Replicating Data

Most clients will choose to replicate the entire dataset (or a portion thereof). To replicate a subset of the data, provide a $filter parameter that describes the subset of data needed. If the entire dataset is desired, provide no $filter parameter at all. By default, the records will be sorted by ModificationTimestamp (in ascending order) and then by the resource Key (in ascending order). The @odata.nextLink may be followed until no results are returned.

After all currently available records have been returned, following the final @odata.nextLink will result in an empty response with no @odata.nextLink. This indicates that replication has reached the end of the currently available data. The @odata.nextLink used for that request may be retained and used during the next polling cycle to retrieve newly available records. If additional data is available, it will be present in the response and a new @odata.nextLink will be provided to continue replication.

The client SHOULD provide a $top parameter to limit the number of records returned in each response. If at any time a page contains fewer records than the value of the $top parameter provided, the client SHOULD defer following the @odata.nextLink until its next polling interval. Doing so will prevent unnecessary API calls and transaction costs.