Apache Cassandra vs HPE Ezmeral Data Fabric

Apache Cassandra is a free, open-source NoSQL distributed database for mission-critical data and large active datasets. Its masterless architecture gives each node the same functionality. The project says read and write throughput can grow linearly as machines are added, without downtime or application interruption. Cassandra can run on premises, with one cloud provider, across multiple providers, or in a combination of those environments. It supports replication across data centers and configurable consistency levels for queries, along with CQL, transactions, and horizontal scaling. The project lists audit logging for DML, DDL, and DCL activity, and fqltool for recording and replaying production workloads. Drivers are available for Java, Go, Python, and Node.js; downloads include tarballs and packages, and installation guidance covers Debian and RPM. The ecosystem includes tools and connectors for Apache Kafka, Apache Flink, Apache Spark, and Kubernetes. The project identifies version 5.0 as the latest GA release and says it is maintained until the 8.0 release. Cassandra is self-managed, and listed third-party ecosystem projects are not supported by its community.

HPE Ezmeral Data Fabric consolidates data from on-premises systems, multiple clouds and edge deployments into one logical store. It supports files, object data, NoSQL databases and streams, with transaction support, strong consistency, multi-region support and horizontal scaling. Application developers can use standard APIs, languages and protocols. Its object store provides S3 API access, WORM compliance, events delivered to data streams and compatibility with open-source AWS S3 and Minio SDKs. Analytics integrations include Apache Spark, Apache Drill, Presto and S3 Select, while certified open-source tools and engines can run directly on the fabric for in-place analytics. Security capabilities include FIPS 140-2 Level 1 support when FIPS mode is enabled and dynamic masking of designated fields in JSON queries. A community edition is available, and an enterprise 30-day trial license can be requested. The community edition has limited nodes, no high availability or technical support, lacks advanced storage options such as snapshots and global replication, and is not intended for production. Pricing is on request; as-a-service monthly charges depend on storage use and selected tier.