StorMagic beefed up its SvSAN hyper-converged software with performance and manageability upgrades, including caching and auto-tiering across solid-state drives, hard disk drives and memory.
StorMagic SvSAN works with VMware and Microsoft Hyper-V hypervisors, and it has expanded its presence in the market through partnerships, such as its deal with Lenovo that targets the low end with a two-node configuration that starts at $5,000. SvSAN is also certified to run with Cisco Unified Computing System E-Series and C-Series servers in two-node configurations. Most of StorMagic's hyper-converged competitors require a three-node minimum for high availability.
The enhanced SvSAN includes an SSD-based read cache to speed up reads to go with its previously supported write cache on flash. The latest version of StorMagic SvSAN also includes memory-based read caching for faster reads. The software uses an algorithm that monitors patterns of frequently accessed data and moves those workloads into the memory cache to speed up those workloads.
The memory-based read cache has two modes for hyper-converged storage that include data pinning and read-ahead capability to speed up performance. Administrators can use the data pinning to target identical or similar data workflows and move those blocks to memory when performing repeat operations in a hyper-converged storage infrastructure.
The read-ahead mode also uses an algorithm to monitor and detect streams of sequential reads from the disk. Read data within that stream can be prefetched and put into memory to improve workload performance.
"[With data pinning], you can essentially record when an event happens, so it captures references on disk," said Luke Pruen, StorMagic's director of technical services. "The data pinning and read ahead plays to more specific use cases, such as analytics, pages, files, databases and VDI [virtual desktop infrastructure] boot storms."
The StorMagic SvSAN's new intelligent auto-tiering places data in memory, on SSDs or on HDDs based on how frequently it is accessed. Administrators can size the memory capacity based on requirements. The most frequently accessed data goes to RAM, with the next most frequently accessed data going on SSDs and cold data on HDDs.
"The key point of intelligent auto-tiering is we are playing to all the strengths of memory, SSDs and spinning disk drives," Pruen said.
A new SvSAN PowerShell auto-script generation feature for the virtual SAN can reproduce configurations without going through the wizard. A new graphically formatted performance statistic tool can be used to better understand workload performance.
StorMagic SvSAN customers can add capacity to existing servers or grow the SvSAN cluster without interrupting availability.
Scott Sinclair, a senior analyst at Enterprise Strategy Group Inc., in Milford, Mass., said StorMagic's newest features are consistent with what other vendors in this market are providing, with the exception of memory-based read cache data pinning.
"They are differentiated on their ability to provide a cost-effective solution that only requires two-nodes, rather than three," Sinclair said.
As bigger players jump in, hyper-converged vendors add features
Why hyper-converged system performance is difficult to guarantee
Essential considerations for converged infrastructure selection