VMworld: Hyperconvergence With NVMe by Scale Computing

Scale Computing, Inc. announced at VMworld 2017 unprecedented results in hyperconverged VM IO performance using NVMe.

Click to enlarge

The lab tests have achieved mean IO latencies as low as twenty microseconds delivered to a guest VM. The architecture of the company’s software-defined storage engine (SCRIBE) coupled with highly responsive NVMe drives enables this performance.

In the coming months, the company will be introducing HyperCore-Direct, a NVMe optimised hyperconverged solution for data intensive environments and customised hardware and network configurations. With the low latency of SCRIBE, HyperCore-Direct and NVMe, the firm will provide optimal performance for both data intensive and legacy workloads.

Legacy software, still in use with many organisations, is not optimised for modern, multi-core CPUs, while new data intensive applications require massive amounts of processing that can waste valuable time. Improving the performance of both of these types of workloads can be challenging, and low storage latency can provide a huge advantage.

Click to enlarge

“Scale’s HC3 product suite has long been a leader in the hyperconverged infrastructure space,” said Phil White, CTO, Scale Computing. “We now continue that leadership by demonstrating the first hyperconverged solution to include NVMe support for maximum performance combined with the availability and scalability that HC3 is known for.”

Click to enlarge

Testing of HyperCore-Direct has shown that applications running within VMs can expect near bare-metal NVMe IO performance from mirrored, highly-available software-defined storage. An example four-node HyperCore-Direct configuration with 100% NAND-based media is able to provide mean IO latencies of 150 microseconds on a mixed (90% read / 10% write) random workload to 24 VMs (at a total 240,000 IO/s). When more responsive media is utilised, the same workload achieves mean latencies of 65 microseconds. The four-node NAND-based system has also demonstrated over 2.6 million IO/s with under 290 microseconds mean latency as delivered to a virtualised benchmark application.

Click to enlarge

“Scale Computing’s new HyperCore-Direct solution brings the ultra-low latency performance of NVMe to the masses,” said Howard Marks, chief scientist, DeepStorage, LLC. “Less than a year ago users requiring the sub-500Ķs latency and low overhead NVMe promised were limited to NVMe SSDs in servers or tier-0 systems that had no data services. Scale’s now delivering that performance on a hyperconverged platform with snapshots, and the other services enterprise applications require.“

“Scale Computing’s solution using Intel Optane SSD P4800X is approaching the 10Ķs or lower bare metal latency for classic virtualisation,” said Frank Ober,Optane technology solution architect, Intel Corp.