If you consider how JSON documents are being stored within MapRDB, its possible to store a B-Tree (or a derivative ) in a MapR table.
Essentially you'd have some sort of structure, a node, that stores the value, and then a set of arrays. (e.g Array Left for the ids of the attached nodes to the left, and Array Right for the nodes that attach to the right.) If this were a binary tree, then there would be at most one node to the right, and one node to the left. Or the empty set. If there are no nodes, then you have a leaf.
The idea is to create a basic structure that would support storing indexes in a secondary table which could be used to reduce the level of effort when attempting to query data within MapRDB.
This would allow for easy secondary indexing within MapRDB and could be associated with the base table.