Crippling CPU usage for upserts

[tsturzl]

We use upserts to update data pulled from our API for offline storage in a electron client app. However, we are upserting a collection of thousands of relatively small documents and I'm seeing the CPU usage spike for a long duration while upserting this data into an nedb database. It becomes so intense on the CPU that the entire application stops responding as a result(sometimes crashing the chrome dev tools). It seems to only happen starting with this collection(its probably the largest) and then every collection after that seems to just slow to a crawl and consume 100% CPU for several minutes.

So additional info the upsert query is leveraging indices, because we are querying the document id for the upsert. We do a single upsert at a time for each document, because as far as I'm aware there is no means of bulk upserts. After running a CPU profile I've found that the most time is spent indexing into an AVLTree(BST), so I feel it may have something to do with indexing.

[tsturzl]

I can upload the CPU profile if they'd be helpful