Why do multikey indexes need to fetch documents after the index scan?

Multikey indexes are specifically designed to handle arrays within documents in MongoDB. When a document contains an array, each entry in that array generates a separate index entry. This means that the index has to create and store individual entries for each value in the array, which allows MongoDB to quickly locate these values during queries.

However, since the index contains references to individual array values rather than the entire document, after the initial index scan, MongoDB still needs to fetch the actual documents from the collection to retrieve all the associated data. This is a necessary step because the index alone does not hold all the information about the documents; it merely points to individual values. Therefore, even though the index scan identifies which array values match the query criteria, the actual document containing those values must be retrieved, resulting in the need for a subsequent fetch.

This mechanism enables MongoDB to efficiently manage and query arrays, but it does mean that there's an additional step involved whenever multikey indexes are utilized, as opposed to single-value indexes, which may not require such document-fetching processes.