Whilst this is optimal for small collections, performance with large collections is greatly reduced since looping through a large dataset and sending each update operation per request to the server incurs a computational penalty.

The Bulk() API comes to the rescue and greatly improves performance since write operations are sent to the server only once in bulk. Efficiency is achieved since the method does not send every write request to the server (as with the current update statement within the forEach() loop) but just once in every 1000 requests, thus making
updates more efficient and quicker than currently is.

Using the same concept above with the forEach() loop to create the batches, we can update the collection in bulk as follows. In this demonstration the Bulk() API available in MongoDB versions  >= 2.6 and < 3.2 uses the initializeUnorderedBulkOp()  method to execute in parallel, as well as in a nondeterministic order, the write
operations in the batches.

It updates all the documents in the clients collection by changing the salary and dob fields to numerical and datetime values respectively: