Similar projects and alternatives to Oak
MapDB provides concurrent Maps, Sets and Queues backed by disk storage or off-heap-memory. It is a fast and easy to use embedded Java database engine.
Replicate your Key Value Store across your network, with consistency, persistance and performance.
Access the most powerful time series database as a service. Ingest, store, & analyze all types of time series data in a fully-managed, purpose-built database. Keep data forever with low-cost storage and superior data compression.
Lasher is an embeddable key-value store written in Java.
Java port of a concurrent trie hash map implementation from the Scala collections library
A gulp of low latency Java
JDK main-line development https://openjdk.org/projects/jdk
FST: fast java serialization drop in-replacement
SaaSHub - Software Alternatives and Reviews. SaaSHub helps you find the best software and product alternatives
Java large off heap cache
Oak reviews and mentions
JEP draft: 64 bit object headers
4 projects | news.ycombinator.com | 11 Oct 2022
Another to add to your collection, https://github.com/yahoo/Oak
Solution for hash-map with >100M values
7 projects | reddit.com/r/java | 21 Dec 2020
Consider using an database (e.g. H2 embedded, redis) with an on-heap cache (e.g. Caffeine). Since you say it is a Zipfian distribution, the cache should absorb most of the requests. For an off-heap hashtable, you might try Oak as it is likely a faster implementation.
yahoo/Oak is an open source project licensed under Apache License 2.0 which is an OSI approved license.