While Hazelcast has run performance benchmarks against Oracle Coherence, the Oracle Technology Network License does not permit us to “disclose results of any program benchmark tests without Oracle’s prior consent.”
However, most people comparing Hazelcast IMDG® with Coherence will actually want to run a benchmark. With this in mind, we have a benchmark suite which we can share with you here.
While Hazelcast and Coherence share many similarities as true in-memory data grids, Hazelcast offers advantages that Coherence users should note.
Architecture
Hazelcast includes features not available in Coherence, including support for multimaps, ring buffers, HyperLogLog, distributed atomic longs, distributed atomic refs, and distributed semaphores.
Languages
In addition to Java, C++, and C# (also supported by Coherence), Hazelcast supports Python, Node.js, Scala, and Go.
Performance Optimizations
Hazelcast supports the near-cache and pipelining capabilities to boost performance.
Reduced Downtime
Hazelcast supports reduced downtime via hot restart and rolling upgrades.
Scale
Hazelcast scales up by storing data in off-heap memory for higher capacity via the High Density Memory Store. Coherence relies on spilling over to disk, and requires running multiple server instances with a very small heap to mitigate garbage collection pauses.
Cloud Support
Hazelcast is a cloud-native in-memory data grid that can be deployed anywhere. Hazelcast provides a lightweight technology in the form of a small JAR file that allows embedding in any Java application. Coherence deployment options are limited.
