Scale Computing Takes Midmarket Hyper-Converged Attack To VMworld; CEO Says New VM Latency Benchmark Is Legacy App 'Game Changer'

Scale Computing – which has built a robust business attacking VMware's licensing model with its hyper-converged virtualization appliance – used VMworld 2017 to showcase its latest breakthrough: a new virtual machine benchmark that delivers what it is calling unprecedented legacy application performance.

The Indianapolis-based Scale is showcasing the performance breakthrough in the heart of its primary competitor's biggest annual event. Specifically, Scale is demonstrating the first hyper-converged solution to include non-volatile memory with NVMe Express on its HC3 appliance.

[Related: CRN Exclusive: Scale Computing CEO On Attacking VMware's Virtualization Licensing Model And In The Process Saving Customers $32M]

That combination has allowed the company to push the technology envelope delivering what it called I/O latencies as low as 20 microseconds for a guest virtual machine.

Scale Computing CEO Jeff Ready said the breakthrough is resonating with customers who are suddenly seeing unprecedented performance of what they thought were applications that would have to be completely re-architected for the cloud era.

Instead, those customers can now take advantage of Scale Computing's HyperCore Direct stack to completely transform their business. "This is a huge legacy application game changer," said Ready in an interview with CRN. "This is massive in terms of what it means for legacy applications. All those companies that have been nursing legacy applications no longer have to rearchitect those solutions."

Traditional all-flash storage run with a latency of two to four milliseconds in comparison of .02 milliseconds with the Scale solution, said Ready. "This is 100 times faster," he said. "What our software innovation has done has exposed just how big a game changer NVMe is."

In the case of a defense contractor that tested the HyperCore solution on a legacy application, the Scale solution reduced the time it took that application to run a report from eight hours to a mind boggling four minutes, said Ready. The customer initially thought the Scale solution was "broken," said Ready. "Literally, the customer assumed it didn't work. When you go from eight hours to a few minutes, you are talking about kicking the can down the road a long way."

Another example: a Windows 7 virtual machine was getting one tenth of a millisecond of latency with 2.5 Gbytes per second of throughput, said Ready. "My joke is that is probably the fastest Windows 7 has ever run," he said. "We picked Windows 7 intentionally because it has been ridiculed for its performance. We are showing its performance smoking!"

Ready said the NVMe low latency is just one more example of how the company is beating bigger competitors in the midmarket with enterprise class performance at a small-to-medium business (SMB) price tag. "We may be the David in the David-Goliath story, but we have some great technology we are putting to good use," he said. "This is awesome."

Ready said the Scale Computing offering is in sharp contrast to enterprise upgrades from other vendors that simply do not deliver the performance customers were expecting. "So often we see customers buy the latest hardware to run old apps and they are consistently disappointed," he said. "Our software stack changes that."

In the coming months, Scale Computing will be working with its 300 partners to implement the NVMe breakthrough on legacy applications via its HyperCore-Direct stack. "We are going to be looking at helping our partners fine tune the back end storage to optimize HyperCore Direct for the breadth and depth of the legacy applications out there from Windows 7 to SAP," he said.

The Scale breakthrough has even attracted the attention of Intel. Frank Ober, an Intel technology solutions architect, said in a press release that the Scale offering based on the Intel Optane P4800X solid state drive approaches the "10µs or lower bare metal latency"  for classic virtualization.

Michael Goldstein, CEO of LAN Infotech, a Fort Lauderdale, Fla., solution provider, said he is anxious to look at the Scale Computing offering in the wake of the legacy application virtualization performance breakthrough.

"We're always looking for the fastest performing solution that brings enterprise performance to our customers at SMB prices," he said. "If Intel is wowed by Scale's applications performance it's a big, big deal. Good for Scale for pushing the technology envelope. This kind of performance is unbelievable if you really start to think about it."

The performance breakthrough comes after Scale earlier this year rolled out its first multi-tier channel program as part of a massive midmarket offensive. Scale has prospered by helping small- and midsized business customers avoid VMware licensing fees. Scale shipped 1,600 appliances in 2016, saving customers what it estimates is about $32 million in VMware licensing costs.

Howard Marks, a highly respected storage guru and chief scientist at DeepStorage, LLC, an independent storage testing lab, praised Scale for getting out of the NVMe gate faster than competitors with performance that bests even highly touted NVMe network and fabric startups. "Companies like E8 (Storage) and Excelero are delivering that performance but without the data services," he said.

Marks said he sees Scale NVMe offering to appeal to midmarket customers with aggressive entry level pricing. "Scale is out early and at their price point not at enterprise price points," he said."This brings the performance of NVMe to their set of customers and price points. To do NVMe from anybody else would be a substantially bigger investment."

Marks expects the Scale solution to take its toll on VMware in the midmarket. "I don't see it as a threat to VMware in the core data center, but it is a huge threat to VMware in the remote office, branch office SME (small medium enterprise) space," he said. "It is biting around the edges rather than attacking VMware at the heart of their market."

Read more articles on: