Just like object queries, an aggregate query whose range or shards parameter selects more than one shard is called a multi-shard query. Because each shard is analogous to a self-contained database, an object with a given ID may be inserted into the same table in multiple shards. Even though objects with the same ID could be considered duplicates, they are considered separate objects for aggregate queries. A COUNT(*) query, for example, will count all selected objects in all shards, ignoring duplicate object IDs.Similarly, when an aggregate query includes unmerged batches by adding the option &uncommitted=true, objects in the batch “mini shards” are treated as separate objects. Since the updates in unmerged batches have not been applied, duplicate object IDs in the update batches and the “live” shard are also considered separate objects. If an update batch deletes an object, the object in the “live” shard will still be present.