Performing a deep $graphLookup of a large graph can result in significant memory consumption due to the retrieval of unnecessary fields.

For example, we have a collection of documents that have:

• id
• parentId
• content

If I want to retrieve all the IDs of a specific graph, the content property is irrelevant.

If the content property is significantly large it can cause "$graphLookup reached maximum memory consumption" errors.  By using a projection to ignore this field within a $graphLookup, memory consumption could be significantly reduced.

Instead, I think we'll have to create a simplified view of the collection and use that to perform the graph lookup, unless there are better suggestions?