MemVerge's Memory Machine is a software package designed for applications that involve memory-intensive computations running on Linux-based servers. Memory Machine manages volatile and persistent memory and provides an array of services made possible by nearly instant in-memory snapshots.
To achieve these benefits, MemVerge Memory Machine adds a software layer that virtualizes system memory and interacts with the runtime memory management functions native to Linux. The internal architecture of Memory Machine assumes that system memory is partitioned into two tiers. The first tier is always composed of DRAM; the second tier may be composed of DRAM or persistent memory. Memory Machine does not require persistent memory. If persistent memory is not installed and persistence is required, then access to storage media (such as SSD) will be necessary.
Memory-intensive applications are used in business-critical environments and so the effect of Memory Machine on overall system availability and performance is important.The business impact of deploying Memory Machine can be significant, ranging from decreasing run times to providing high availability (HA) to reducing total cost of ownership (TCO). Memory Machine future-proofs applications from changes in memory technology including new CPU-to-memory interconnects such as CXL.
Memory Machine is part of the trend towards software-defined infrastructure in modern data center architectures. More information at: https://www.memverge.com