Today, Atlantis Computing moves into the hardware market with a new hyperconverged solution, HyperScale. HyperScale is based on the company’s flagship product, USX. Technically, this solution is not a revolution, but it is an evolution on Atlantis Computing’s part. This is the first time it has delivered an end-to-end bespoke solution that tightly couples certified hardware with its flagship USX product. More to the point, unlike most new entrants into this space, Atlantis has entered straight in with a full product set, multiple-hypervisor support, and three OEM deals. This is in addition to its own Supermicro-based in-house appliance. What’s more, HyperScale has a starting price that does not set your teeth on edge.

I attended an early briefing on this new product, and it sounds impressive. Many will know I have had a love-hate relationship with Atlantis. Its first product, ILIO, had some issues in the early days. This is not the case with USX. USX is a solid product. What interests me about HyperScale is not the fact that Atlantis has released an appliance-based product. Rather, it is that out of the box, Atlantis is supporting multiple hypervisors—currently vSphere and XenServer. Further, it already has a solid OEM ecosystem, with offerings on day one from HP, Cisco, and Lenovo.

This is quite impressive for a day-zero release. Currently, I have no detailed information about the OEM offerings, but the in-house version utilizes a four-node Supermicro TwinPro2. This is a nice SLED device. These devices can take 1 TB of RAM across the four nodes in each chassis, although Atlantis Computing’s CX-12 model comes with 256 to 512 GB across four of the nodes, and its CX-24 utilizes between 384 and 512 GB. Networking is provided by two 10 GbE and two 1 GbE devices. As with all SLED devices, each node has access to a number of storage bays. The CX-12 has four 400 GB SSD devices installed, and the CX-24 utilizes four 800 GB devices. On the face of it, this might seem a little sparse on storage, but as these devices are utilized as a USX hybrid volume, this is not that much of an issue.

In fact, I would not be surprised to find that those remaining slots could be utilized to provide a VSAN datastore to ESXi; USX is one of the few flash acceleration products in the market supported by VMware when coupled with VSAN.

Atlantis has priced this aggressively. An entry-level four-node CX-12 device starts at $78,000. I am confident that prices do not include costs relating to the hypervisor level. But even so, that is an aggressive price point. It includes all hardware costs as well as a 24/7 three-year single-point telephone and hypervisor support contract, with four-hour response as standard.

How does this compare with Nutanix, the current market leader in this space? In an apples-to-apples comparison, the NX-3000 list price is circa $420K, and the CX-24 list price is $145K. That is a savings of 65% on list.

Performance-wise, on paper this should eat the competition for breakfast. Unlike Nutanix, SimpliVity, or even EVO:RAIL, HyperScale uses RAM for its acceleration, rather than the flash layer.

All in all, this is a very interesting entrant to the hyperconverged market space. I look forward to seeing it up close and personal when I evaluate possible solutions for my customers.