Based on Intel’s Gen 10 NUC, the small form factor HE150 has the same software stack as Scale’s HC3 appliance, but is aimed at even smaller sites without even a small server closet, as it takes up only the space as three stacked smartphones.
The new Scale HE150
Hyperconverged infrastructure [HCI] vendor Scale Computing, which has historically had a focus on the edge that predates the common use of that term, is today introducing the HE150, a new small form factor version of its HC3 appliance. The all-flash HE150 has the same software stack, and thus all the functionality of the HC3. Like the HC3, the HE150 is a 3-node High Availability cluster. However, it is designed for deployments which lack the room, or the budget, for a larger VSA appliance.
“The HE150 is us capitalizing on something that no one else can do – bring the HCI stack to the smallest of the small at a price no one else can even get near at $5,000,” said Alan Conboy, Office of the CTO at Scale. “For coffee chains, and drugstore companies, having a $20-30,000 cluster at each site isn’t appropriate.”
Scale already has a strong presence in the retail edge market, including a relationship with European grocery giant Delhaize, that has been based on Scale’s HC3 3-node solution. The HE150 is also a 3-node offering, but is aimed at even smaller deployments in these edge use cases.
“There are environments beyond the retail edge like large coffee shop chains, and distributed locations like oil and gas where there is a need for something even smaller – where there is no back office at all,” Conboy said. “In these stores, they don’t have a place to put anything. They need something that is small, self-healing and novice- friendly from an IT perspective.”
The HE150 is designed for the ultra-space sensitive.
“Our footprint in RAM, disk and core count has always been a differentiator,” Conboy stated. “The HE150 puts our money where our mouth is. The footprint with the NUC goes down to a GB of RAM a node with no loss in feature functionality. It means using the same code base and the same stack as the HC3. However, it also means that we take up the size of a couple smartphones, that fit on a shelf under the register.” The form factor is 117x112x38 millimetres.
The HE150 is based on Intel’s 10th generation ‘Frost Canyon’ NUC. It is designed for deployments with limited storage needs, with 250GB, 500GB, 1TB, 2TB NVMe options within a single disk. Conboy said that it’s also not something that competitors can replicate using the same NUC,
“The magic isn’t really in the Intel product itself,” he said. “NUCs have been around close to 11 years. It’s the efficiency of the stack running on the NUC. Gen 10 NUC comes with 32GB of RAM and 8 cores, if you include hyperthreading. There are two categories of HCI. One is Virtual SAN Appliance-based. Those VSAs start their RAM consumption at 32 GB. If your edge platform only has 32 GB like this, these appliances consume all the resources just to boot themselves up – to say nothing of running workloads.
“The other category, which Scale started and VMware jumped into, is kernel-based,” Conboy added. “There are three players in kernel. Microsoft, with HyperV, is inefficient. VMware’s vSAN in VxRail has a minimum of two drives in every node, while Scale’s minimum requirement is a single drive. We are the one player who can physically run their stack and leave most resources for running virtual machines. Because we have built our stack to be so viciously efficient, we can go into places they just can’t – at a price it would have cost just for a server.”
Conboy emphasized that while the HE150 just has a single drive, its self-healing capabilities will keep the device running when a drive inevitably breaks down, something that is critical in its target use cases where no IT will be present.
“With the Scale HC3, if a drive dies there are multiple drives in each node,” he said. “There aren’t in the HE150. It’s about resource optimization, so there is a single drive. However, the workloads are in a minimum of two separate drives on two separate nodes. So when a node dies, the state machines on each of the other nodes recognize the node has gone away, offline the bad one, and have a full copy of the data blocks spread across their drive. They pull them out without the end user having to do anything.”
Scale is emphasizing that the HE150 can be deployed in under an hour, and easily managed remotely.
“We have had these in POCs for several months with several very large companies,” Conboy noted. “Some have jumped the gun and already moved them into production. It allows them to modernize their edge infrastructure, while fitting into places where legacy products couldn’t go.”
The HE150 appliance is available now.
