The experience of cloud computing should be seamless, ephemeral, elastic and scalable. When it works well, it just magically happens. If on-premises computing is to offer the same experience, those attributes need to be baked into the architecture of your infrastructure components. And that’s the HyperCloud difference. It’s how you get the ‘it just works’ cloud experience on-premises.

HyperCloud, private, on-premises compute.

  • Automatically deploy and scale resources according to workload
  • Radically reduced latency, especially with data on-premises
  • Protect your processes and outputs as critical IP
  • Update and manage as a single entity from a one-pane dashboard
  • Burst compute at scale with a HyperCloud/public cloud hybrid

Each HyperCloud compute node is a seamless addition to your private HyperCloud. The moment it’s plugged in, it gains peer recognition through the network, and updates and instantly expands fleet intelligence, capacity, and more.

No more patchwork is needed trying to get a Franken-system of mismatched components operating smoothly. HyperCloud is one system. It’s simple to run, highly flexible and easy to live with—by design.

Compute at a glance

Born as cloud, wakes as cloud

HyperCloud compute nodes were always private cloud nodes. They didn’t evolve from legacy server technology.

Upon power-up, they integrate into the HyperCloud fleet. Once they find their HyperCloud, they download the latest software, and are immediately added to the compute pool. Additionally, they contribute to private cloud’s distributed fleet intelligence.

Ephemeral

Deploy and scale compute resources according to workload requirements using our cloud CLI, GUI, or APIs. To replace a physical compute node, simply remove and swap it out. HyperCloud automatically redistributes your workloads.

Elastic

Deploy compute resources as you need them, and then scale up or down as your workloads require (using our cloud GUI, CLI, or through APIs).

Provenance

HyperCloud has security built in from the inside out. Every line of code that powers our HyperCloud compute nodes not written by SoftIron is compiled from source. That provides a complete, auditable chain—something unrivaled by any cloud-washed alternative.

Simple

HyperCloud simplifies compute management akin to public cloud. Use our CLI, GUI, or APIs. Updates are applied universally, not individually.

Efficient

SoftIron’s “less is more” build philosophy means our compute nodes consume much less power and require much less cooling than cloud-washed alternatives.

Specialized compute

Quick, what do these workloads have in common?

  • AI/ML
  • Animation rendering
  • Transcoding video

Answer: Almost nothing. So why would you run all three workloads on the same generic compute platform?

Compute node technical specifications

Compute node technical specifications
AMD EPYC™ Embedded 3rd Gen AMD EPYC™
CPU 8‑Core (16 Thread) – 16‑Core (32 Thread) 16‑Core (32 Thread) – 64‑Core (128 Thread)
Memory 128 GB – 1,024 GB 256 GB – 2,048 GB
Interconnect 2 × 25 GbE SFP+ 2 × 25 GbE SFP+
Power supply 2 × 110 – 240 VAC (C14) 2 × 110 – 240 VAC (C14)
Typical power consumption 110 – 165 W TBA
Typical BTU/hr 375 – 563 BTU/hr TBA
Typical weight 11.5 kg / 25.5 lbs 12.0 kg / 26.5 lbs
Dimensions H: 1 RU
W: 485 mm / 19.1 in
L: 637 mm / 30.25 in
H: 1 RU
W: 485 mm / 19.1 in
L: 637 mm / 30.25 in

HyperCloud in action

Capturing flight data at the edge

The United States Air Force 96th Test Wing stores 5.4PB per aircraft per year of audio, video, and telemetry data from test flights, all on a private cloud built using HyperCloud by SoftIron.

Discover more applications

HyperCloud lets you deploy, manage, and curate your on-premises storage with the advantage you’re used to in the public cloud.
HyperCloud provides the single, unified infrastructure needed to run fast, efficient, and portable cloud-native workloads on your private cloud.
HyperCloud provides a unified hardware and software solution that lets you run virtual workloads at typically half the price and with a fraction of the complexity.
Back to Top