Want to call out that while there are presumably correctness issues to this when it comes to sharding, most uses of this sort of pattern I assume are to ultimately update all documents in a large collection without overwhelming the cluster.
If you have 1 billion documents (frankly lower numbers this still holds) and I want to update each one to add a field (or remove a field).
Right now there is not an easy way to slowly migrate these without doing something that feels way less performant than what MongoDB could do internally.
Want to call out that while there are presumably correctness issues to this when it comes to sharding, most uses of this sort of pattern I assume are to ultimately update all documents in a large collection without overwhelming the cluster.
If you have 1 billion documents (frankly lower numbers this still holds) and I want to update each one to add a field (or remove a field).
Right now there is not an easy way to slowly migrate these without doing something that feels way less performant than what MongoDB could do internally.