You need to scale out SSAS. What should you do?

Backup and restore is a disaster recovery and migration technique - it is slower and more manual than synchronization, and is not designed as a recurring scale-out mechanism.

Linked objects in SSAS allow referencing metadata objects from another database but do not replicate cube data to another server, so query load is not actually distributed.

Processing the cube independently on both servers each day risks data inconsistencies between nodes if source data changes mid-process, and it doubles processing resource costs without using the proper synchronization pattern.

SSAS supports a scale-out architecture where one server handles resource-intensive cube processing and the completed database is then synchronized to one or more read-only query servers using the Synchronize Database command or XMLA. This isolates processing overhead from user query traffic, ensuring query servers remain responsive. Daily synchronization keeps query servers current while avoiding the cost of redundant reprocessing on each node.