Hazelcast, an open-source in-memory computing platform, announced a partnership with Intel to support Intel’s Optane DC Persistent Memory—a storage-class memory (SCM) solution that allows 512 GB of Optane storage (3D XPoint) to be utilized as RAM.
In-memory computing solutions such as Hazelcast are useful for high-performance or high-transaction databases and AI model training, and other workloads that require simultaneous management of multiple data streams, efficient processing of large data sets, or storing larger working sets in memory.
SEE: Special report: Prepare for serverless computing (free PDF) (TechRepublic)
Optane is not as fast as DRAM, the use of Optane as RAM provides for substantially faster transaction speeds, as writes do not need to be immediately pushed through PCIe-linked solid-state storage, eliminating a significant bottleneck. Likewise, the non-volatile nature of SSDs is preserved in Optane DIMMs, decreasing downtime across reboots.
Intel’s Optane DC Persistent Memory offering has a great deal of potential for enterprise compute, though Intel has struggled to convince enterprise buyers of the merits of Optane. To fully utilize the technology, some application-level reengineering is necessary. With Hazelcast joining the fold, current users of Hazelcast—including JPMorgan Chase, Verizon, HP, and PayPal, among others—could find themselves with reason to adopt the technology.
The rapidly-falling prices of DRAM, however, may have played a contributing factor in the hurdles Intel faces in their pitches to enterprise buyers. Optane holds an advantage in density, though lower costs for standard DRAM allow for drop-in performance upgrades, without the re-engineering needs that accompany adoption of Optane.
Hazelcast is the latest group to join the Optane club, with Dell EMC adding support to PowerMax earlier this month. Support for Optane DIMMs is also present in Redis, while HP sees memory-driven computing generally as the future for enterprise computing.
For more, check out “Researchers are performing integer factorization using modified MRAM” and “IBM z15: Multicloud makes the case for why mainframes still matter” at TechRepublic.