SoftIron's Technology
SoftIron resolves the CIO’s dilemma. by delivering a public-cloud experience at the network edge, with all the advantages and promise, but without humans in the control plane, and without the high cost and limitations.
Legacy IT deployments (including clouds) are built from disparate parts, often from many vendors. SoftIron, in contrast, revolutionizes the IT equation by offering a “pourable” single technology that embodies and enables the virtualization of everything.
At SoftIron, we “solve for” a different technological problem. We don’t assume scale, we don’t assume connectivity, and we don’t even assume skills or people. We believe that enterprise IT deployments should be as simple as home computers.
Stateless Hardware
True statelessness implies that components are interchangeable because their identity and “value” are defined elsewhere. In practice, modern data centers have rarely been stateless from the “power socket and Ethernet cable up." “Statelessness,” in this context, usually sits atop massive state stacks of servers and software. A core principle of SoftIron is that rack hardware should have no special significance except as compute, storage, acceleration, or network resources. Our hardware does not create topology or dependency constraints. There are no “sacred” machines. Instead, servers are simply vessels for resources, and act conjointly as a sea of resources without limits.
Virtualize Everything
Public cloud reflects how most organizations want to consume IT: as on-demand resources, not as infrastructure requiring dedicated, complex teams to build, maintain, and operate. With SoftIron, those resources are virtualized and immediately available for direct use. The challenge is to extend this model to the underlying on-premises estate, to deliver an “edge-first public-cloud” experience without the operational burden of managing disparate technologies and hardware across lifecycles, upgrades, failures, and integration gaps. This is achieved by virtualizing the data center fleet itself, starting with “stateless hardware.”
Boot from Cloud
Armed with the critical tools of truly stateless rack hardware and a ViRB for fleet management, we can now deliver resources with true public-cloud properties. We begin by booting directly from the cloud, with software, configuration, state, and instructions provided at runtime. Resources are then presented and consumed like a public cloud, while servers remain interchangeable: able to fail, return, upgrade, or be reassigned in real time within a federated resource pool. This model further spans acceleration, storage, and networking.
In this architecture, the hardware estate is defined by resources–not topology, fixed configurations, or “pipes” that must be tuned and maintained. Lifecycle management moves beneath the abstraction. To make this reliable, we boot stateless hardware with a guaranteed, known-good state.
One Binary
At SoftIron, all rack hardware (servers and supporting components) runs a single binary image, not just a common codebase. Booting from the cloud is therefore simple and deterministic because configuration and state are delivered at runtime. Interoperability is guaranteed. In a stateless fleet, swapping, failure, upgrades, and reassignment are all operationally equivalent to a reboot. Interoperability isn’t a separate project; the architecture makes it a built-in assumption.
The remaining question is how a single software image can support modular resources across compute, acceleration, storage, and networking. For SoftIron, the answer lies in a modular design wherein modules connect through a common backplane. The technology is deep, but the purpose is straightforward: deliver cloud-like simplicity, public-cloud properties, and frictionless consumption on premises, and without compromise.
Technology Components
Peers and Scale
Read More about Peers and Scale
Mystic Rose
Read More about Mystic Rose
Topology Transcendence
Read More about Topology TranscendenceEverything Software Defined
“Software-defined” is both industry shorthand and a real class of technologies layered on top of existing stacks. SoftIron applies this model to virtualize all resources, thus permitting multiple, parallel uses of the same underlying asset. This is similar to processor virtualization within a single server.
With SoftIron, resources are attached directly, without configuration. Once allocated to a virtual machine or tenancy, they have native, local availability without an intervening protocol layer. As a result, even legacy applications can consume resources within a software-defined model: scalable, high-performance, dynamically adaptable, and abstracted from hardware failures. See Self-Assembling, Self-Organizing, and Self-Healing.
High Side or Hands Off Management
Disconnected or highly classified environments often have limited access to specialized staff, yet still run mission-critical data and workloads. In these settings, the public-cloud operating model is ideal. SoftIron is designed for disconnection as a baseline assumption, while still preserving its design imperatives without compromise.
Freedom from the Hardware Preoccupation
Organizations should never have to think about hardware, including how a cloud is built. Today, even hyperscalers assemble clouds from multiple vendors’ components across compute, acceleration, storage, and networking, thereby embedding high technical-support and maintenance requirements.
SoftIron is purpose-built to compose clouds as unified systems. It presents resources as a single, cohesive fabric, not as a collection of disparate components. This enables an edge-first public-cloud model: no assumed centralization, no required internet connectivity, no need for hyperscale, and no large operational army!
Self Assembling, Self Organizing, and Self Healing
In a public cloud, administration is largely invisible and consumption is the priority; applications and workloads come first. With SoftIron, this kind of performance is possible because resources are federated into a peer-based pool, unified by the architecture itself. The virtualized environment operates as a single, self-aware system with one consistent mechanism for resource virtualization, so, like public cloud, users can simply consume resources.
SoftIron scales by adding servers incrementally or in bulk. Power and network are connected, and the system self-assembles in a deeply, fully integrated way. As compute, acceleration, storage, and networking resources are added, workloads can be rebalanced and moved, enabling a self-organizing cluster.
The SoftIron platform continuously optimizes toward the outcomes defined by its design imperatives, handling lifecycle operations and day-to-day management. This removes both the operational burden and the risks of human error, inconsistency, and unnecessary cost, thus freeing teams to focus on workloads with an expectation of full system availability.
When exceptions occur (hardware failures or external events), SoftIron maintains its design imperatives, ensuring bits, cycles, and packets are preserved. Stateless, replaceable hardware and a federated peer-resource pool enable resilient operation without immediate human intervention, thereby dramatically reducing downtime and improving outcomes.
Resources Not Boxes
The result is an architecture wherein resources are made available and consumed exactly like in public cloud. Users (and IT departments) don’t need to think about servers, hardware, topology, or the physical layer. This is the bottom-up outcome of the architecture.
Network Holds the State
With SoftIron, all state is stored in the network hardware itself, where it is accessible, distributed, redundant and highly available. SoftIron makes the network what it always needed to be: a service to provide services (or resources).
This is the technology that allows us to bring the public cloud to the edge, to provide an "edge-first public cloud" with public-cloud benefits and none of the compromises.