Hazelcast Platform vs Ristretto

Hazelcast Platform brings a distributed compute engine and an in-memory data store together for real-time and AI-driven applications. It can process streaming data before storage, supporting action on data in motion, and provides fast reads and writes for interactive request/response workloads. SQL processing, stateful processing, event-time windows, and Java and SQL are listed capabilities. WAN Replication synchronizes clusters and supports automatic failover to a secondary cluster for disaster recovery. Security features include JAAS role-based access controls and TLS encryption for data moving between nodes and applications. The platform runs on public and private clouds, is available through AWS, Google Cloud, and Azure marketplaces, and supports hybrid deployment. Its Kafka Connector processes Apache Kafka topic streams and sends processed data back to Kafka. Hazelcast identifies mission-critical, data-intensive AI workloads as a focus, including use in financial services, retail, supply chain, and healthcare. The platform is available for Linux, macOS, and Windows, as well as through an API and self-hosted deployment. A free plan and free 30-day Enterprise trial are listed; the trial requires a requested license key. The annual self-managed subscription includes professional support, with pricing available by request.

Ristretto is a free, open-source Go library for an in-process cache of key-value data, designed for concurrent use, performance and correctness. It keeps data in memory in a single process rather than operating as a distributed cache. TinyLFU admission and SampledLFU eviction are paired to target high hit ratios; applications define item cost, allowing a valuable large item to displace several smaller ones. The project attributes throughput to contention management, batching and eventual consistency. Goroutines can use the cache concurrently with little throughput degradation, according to the project, and optional metrics cover throughput, hit ratios and other statistics. A Set for a new item may be dropped in the Set buffer or admission policy, while updates are guaranteed. Ristretto requires Go 1.21 or later and Go modules. Version 2 supports generics, which the project recommends for new programs. The repository is licensed under Apache License, Version 2.0, and describes the library as production-ready. Its README names Badger and Dgraph as projects using it.