jankotek/mapdb is a repository for an embedded database engine that works with maps, lists, and queues through Java collection interfaces. It runs on either disk storage or off-heap memory (memory outside the Java heap). The published information contains no release version, so the version this repository was released as cannot be confirmed.
The point of MapDB is that familiar collection types and a storage engine are bundled together. A store that can stand in for Map, List, and Queue sits inside the application, and the README explains that off-heap collections use off-heap memory, so they are handled independently of the garbage collector's (GC) timing. Supported items include a multi-level cache with expiration (expire) and disk overflow, transactions, MVCC, and incremental backup. MVCC (multi-version concurrency control) is a method that keeps several versions of the data so that reads and writes run side by side without blocking each other. The license is Apache-2.0, the repository has collected 5054 stars, and it was created on August 17, 2012.
That said, only a feature list is presented. There are no performance figures comparing it with other embedded DBs, and no material showing what off-heap collections actually gain in a real JVM. Of the published scores, the momentum indicator is low at 1.0 out of 10, and no novelty, technology, or quality scores are given. The language labeling also needs checking, because the repository information says Java while the README says it is implemented in Kotlin. It is worth keeping in mind as an embedded storage candidate with a wide feature scope and an Apache-2.0 license, but since it was created on August 17, 2012, the maintenance history and release status should be checked first before choosing it.