PernixData, which specializes in caching data in random access memory and is able to cluster groups of hypervisors and/or servers to do so for big data deployments, said Feb. 25 that it has added compression within the RAM, extending its performance to new levels.
The San Jose, Calif.-based company’s FVP v2.5 platform now features something called Distributed Fault-Tolerant Memory with LZ4 compression algorithms. These fast, automated algorithms do not lose data in the process, PernixData said, and they make processing decisions faster. DFTM-Z is self-monitoring in that it checks data to see if the algorithm should be applied in the first place. A physical host server’s RAM is limited to about 25 percent of the total, but even 25 percent still can play a major role in faster storage processing — especially in large deployments, the company said.
The net result is a fourfold increase in RAM capacity using DFTM-Z, PernixData said. This speeds up a system’s ability to move its virtual machines faster, and speedier processing is always the goal.
Key features of the FVP packages include hypervisor clustering, topology- aware fault tolerance, read and write acceleration, and seamless support for any shared block and file storage systems.
FVP v2.5 also has templates called Intelligent IO profiles. These ostensibly can determine in real time which workloads are best suited for server-side acceleration and which are not. This avoids using up flash and RAM resource pools on non-business critical data, according to the company.
The Intelligent IO Profiling and RBAC features are available now with all editions of FVP v2.5 (Standard, Enterprise, Subscription, VDI and Essentials Plus).
In only a matter of months, PernixData said, hundreds of companies across the globe are now processing about 200,000 virtual workloads daily using a decoupled storage architecture that includes their own legacy hardware and the FVP software, which runs on any server or operating system.