Skip to main content Skip to main content

Hyper-Unity Architectural Overview

What is HyperConverged Infrastructure (HCI)?

HyperConverged Infrastructure (HCI) is a modern approach that streamlines IT infrastructure and management by alleviating traditional burdens associated with multiple independent and proprietary compute, storage and network subsystems. Inherently a “Software-Defined” approach, HCI builds and operates an enterprise datacenter with many of the efficiencies and advantages of a Public Cloud. It is an ideal and fundamental building block for companies looking to design and implement a Private Cloud solution without the old configuration, operation, and management complexities.

Utilizing virtualization and other software-defined methods to provide a single, unified operational interface that can provision, deploy and manage all required IT infrastructure, HCI simplifies the delivery of enterprise-class applications and services. At its core, this is implemented by aggregating standard x86-based servers with direct-attached storage into flexible pools of compute and storage resources,all interconnected with standard networking.

Utilizing HCI, traditional “silos” of (often proprietary) resources can be broken down and replaced by common and standard servers and storage. This enables the user to modularly scale the infrastructure “out” simply and dynamically - as opposed to upgrading and/or replacing entire complex subsystems when limits have been reached.

Rapid Deployment

The Hyper-Unity Platform - What does it do for HCI solutions?

As HCI concepts originated from architectural tenets found in large datacenters and elastic Public Clouds, nearly all implementations have been in Enterprise Datacenter environments, using commercial hardware in controlled surroundings with ample power, cooling and rack space.

This does not bode well for remote and compute-on-the-move applications, which require robust infrastructure and far more efficient form factors specifically designed to operate in harsh conditions such as extreme temperature, dirt, dust, vibration and shock.

Hyper-Unity removes these HCI deployment limitations, providing complete, fully-integrated platforms optimized for leading HCI software and running that software on its proven server/storage product portfolio. Hyper-Unity brings the advantage of HCI to operations on the tactical edge with size, weight, and power (SWaP) optimized, reliability-enhanced, rugged capabilities that also meet a wide range of Military specifications.

Optimized Efficiency 

A Hyper-Unity platform is a cluster comprising of two to sixteen servers with direct attached storage, interconnected with redundant high-speed, low-latency 40/56Gbps Ethernet switches. All “n” nodes are identical, modular building block aggregated together to form a unified cluster of rugged, SWaP optimized, compute, storage and network resources. The HCI software provisions cluster resources into virtual pools that virtual machines can dynamically deploy according to individual workloads. This way the utilization of cluster resources is undeterred by workload types, variances, and numbers, creating optimal efficiency.

Effortless Scaling

Hyper-Unity HCI platforms utilize standard and configurable Themis components that allow cluster nodes to be “scaled up” in performance/CPU cores, RAM, and storage capacity. The entire cluster can also be “scaled out” by simply adding additional nodes to the cluster. Scaling out is remarkably simple, as it is handled dynamically by HCI software without disrupting running applications. Additional capacity provided by the new nodes added to the cluster become immediately available for deployment by the HCI management interface. A simple illustration highlights this capability.

As Hyper-Unity HCI platforms include direct attached solid state drives and redundant high-speed, low-latency 40/56Gbps Ethernet switches, intra-cluster communication is optimized for maximum application performance. This includes data transfers and throughput, rapid application migration and failover operations, and cluster management functions performed by the HCI software.

The Perfect Form and Fit

Hyper-Unity Clusters—Start with what you need, then “grow-as-you go”

Since HCI is inherently software-defined and employs virtualization technology, it runs equally well on multiple Mercury products and form factors, whether it’s a 19” rackmount RES-1U server, a RES High-Density (RES-HD) server, or a modular and shallow 10” deep HDversa server cluster. With our industry-leading SWaP (Size, Weight and Power) optimizations, customers can select a platform with the best form and fit for their deployment environment while defining requisite performance and capacity according to application needs. A Hyper-Unity cluster can be configured to support compute-intensive and I/O-intensive workloads equally well or lean towards one or the other.

Today, Hyper-Unity is qualified and offered with standard RES-XR6-1U, eight drive nodes. These cluster nodes feature dual-socket Intel™ Xeon Scalable Gold CPUs with up to 40 CPU cores per node - resulting in 80 hyperthreaded cores, 1.5TB of RAM and 8 SSD drives with up to 8TB each. This translates to an aggregated 64TB of storage in the cluster pool.

Cluster sizes begin with three nodes and two 12-port switches for fully-redundant, No-Single-Point-of-Failure (NSPF) operation. The base node size consists of dual 16-core processors, 384GB RAM and eight 1TB SSDs. This provides the base three node cluster with 96 CPU cores (192 hyperthreaded cores), 1.15TB RAM and 24TB of raw storage capacity.

All of these resources are available to, and under direct control of the HCI software - which dynamically and optimally provisions them according to executed applications and virtual machine parameters. Current cluster sizes are from three to twelve nodes. Clusters of 16 to 32 nodes utilizing larger switches will be available in the near future. For detailed information and configuration options refer to our Hyper-Unity product page.