A memory virtualization vendor this week announced software that allows x86-based servers to share memory and make a pool of memory available for large data sets that require it.

Called RNAcache, the software is the second package from this company – the first was RNAmessenger, software that speeds processing for low-latency applications.

RNAcache is installed on each x86-based server and virtualizes the memory of each server into a pooled and and shared resource. It provides line speed processing for memory-intensive applications such as predictive analysis and modeling and clustered server environments.

RNAcache works Gigabit Ethernet, 10G Ethernet and InfiniBand environments.

According to company claims, in predictive analytics and modeling environments, the software improved capabilities by 20x; for fast Internet applications, queries were sped up 100x.

In an RNAcache-enabled environment, the pooled memory appears to each server as local memory. Metadata management allows the distribution of location-based cache data to each application node, improving performance and scalability.

RNAcache consists of three components: an RNAdriver that lets nodes read and write data into the memory pool; RNAserver, a dedicated or non-dedicated server that lets nodes contribute memory to the pool; and, RNAmanager which provisions, controls and monitor the memory pool.

At this point, RNAcache runs on only Unix or Linux-based servers. It supports RDMA.

Deni Connor is principal analyst for Storage Strategies NOW and host of both the Masters of Storage and Masters of Servers Solution Centers.