I've only heard of this trick being employed on OLAP scenarios. Is this kind of partitioning also advisable for OLTP workloads?

Depends on the workload. In the past, I've worked on several workflow-based systems that performed lots of OLTP operations to drive live workflows forward, but once a workflow was done the operational data became a lot less interesting. So there it made sense to (say) partition the operational data and table by month, and roll off partitions after 3-6 months.