MapDB combines embedded database engine and Java collections. It is free under Apache 2 license. MapDB is flexible and can be used in many roles:

• Drop-in replacement for Maps, Lists, Queues and other collections.
• Off-heap collections not affected by Garbage Collector
• Multilevel cache with expiration and disk overflow.
• RDBMs replacement with transactions, MVCC, incremental backups etc…
• Local data processing and filtering. MapDB has utilities to process huge quantities of data in reasonable time.

Maven snippet, VERSION is

<dependency>
    <groupId>org.mapdb</groupId>
    <artifactId>mapdb</artifactId>
    <version>VERSION</version>
</dependency>

Hello world:

//import org.mapdb.*
DB db = DBMaker.memoryDB().make();
ConcurrentMap map = db.hashMap("map").make();
map.put("something", "here");

You can continue with quick start or refer to the documentation.

More details.

MapDB is written in Kotlin, you will need IntelliJ Idea.

You can use Gradle to build MapDB.

MapDB is extensively unit-tested. By default, only tiny fraction of all tests are executed, so build finishes under 10 minutes. Full test suite has over million test cases and runs for several hours/days. To run full test suite, set -Dmdbtest=1 VM option.

Longer unit tests might require more memory. Use this to increase heap memory assigned to unit tests: -DtestArgLine="-Xmx3G"

By default unit tests are executed in 3 threads. Thread count is controlled by -DtestThreadCount=3 property

On machine with limited memory you can change fork mode so unit test consume less RAM, but run longer: -DtestReuseForks=false