FEATURES AND SPECIFICATIONS
Features and hardware specifications for deploying StratiSTOR, StratiSERV, and HyperSERV.
STRATISYSTEM - FEATURES AND SPECIFICATIONS - REV 4.10
OPEN STANDARDS PLATFORM
StratiSYSTEM OS is an infrastructure operating system that unifies storage, virtualization, containers, cloud-native services, networking, self-hosted AI inference, HPC acceleration, monitoring, and lifecycle operations on a software-defined foundation. Open standards and hardware choice allow organizations to modernize without replacing every infrastructure layer or becoming dependent on a single proprietary appliance architecture.
StratiSYSTEM is designed for heterogeneous enterprise environments. It can use industry-standard Intel and AMD x86 systems, local or external storage, high-speed Ethernet and specialized fabrics, GPUs and DPUs, and third-party infrastructure where required. Compatible platforms can also use Compute Express Link (CXL) memory expansion. StratiSTOR provides software-defined data services, StratiSERV provides compute and virtualization services, and HyperSERV combines both on the same nodes. The same operational model can extend from compact edge systems to large clustered, multi-site, AI, HPC, and private-cloud environments.
This architecture addresses a central challenge in modern infrastructure: organizations increasingly need to operate traditional virtual machines, Kubernetes, AI workloads, large data services, high-speed networks, and distributed sites at the same time. StratiSYSTEM provides a common deployment and lifecycle layer while preserving the ability to select the hardware, topology, and service model that best fits each workload.
MODERN INFRASTRUCTURE DESIGN PRIORITIES
The following design priorities summarize how StratiSYSTEM is positioned for current enterprise, service-provider, AI, HPC, edge, and private-cloud infrastructure requirements.
| INFRASTRUCTURE PRIORITY | STRATISYSTEM APPROACH |
|---|---|
| Reduce Infrastructure Fragmentation | Consolidates storage, virtualization, containers, Kubernetes, networking, AI/HPC services, monitoring, and lifecycle operations under a common infrastructure operating model. This reduces the number of isolated infrastructure stacks that must be deployed, integrated, patched, and operated independently. |
| Support Mixed Workload Generations | Runs traditional enterprise applications, virtual machines, cloud-native services, AI pipelines, data platforms, VDI, and HPC workloads on the same operating foundation, allowing modernization to occur incrementally rather than through disruptive platform replacement. |
| Increase Data-Plane Performance | Combines high-speed Ethernet, RDMA/RoCE, InfiniBand-oriented fabrics, SR-IOV, DPDK, VPP, multiqueue networking, and hardware offload options to address the packet-processing and east-west traffic demands created by 100GbE-and-faster networks, distributed storage, containers, and AI workloads. |
| Keep Data Close to Compute | Brings StratiSTOR data services, Lustre, virtualization, Kubernetes, and GPU-enabled compute together so large datasets can remain close to the workloads that consume them, reducing unnecessary movement and helping address data-gravity challenges. |
| Operate AI and GPU Infrastructure | Combines supported CPU and GPU inference, multi-model local AI services, GPU passthrough, compatible vGPU technologies, GPU monitoring, high-throughput storage, and high-speed networking. Optional retrieval, web search, and memory services extend private AI workflows without requiring a public-cloud inference endpoint. |
| Improve Cyber Resilience and Control | Combines role-aware administration, isolation, at-rest and in-transit protection where configured, Secure Boot, virtual TPM, offline operations, and an optional post-quantum secure fabric for site connectivity. Recovery workflows include VM backups, replication, and selected guest-file restoration. |
| Design for Failure and Multi-Site Availability | Supports fault-domain-aware clustering, self-healing data services, high availability, replication, erasure coding, backup, stretch architectures, and qualified synchronous multi-site protection for workloads that require an RPO of zero. |
| Modernize Without Hardware Lock-In | Uses industry-standard x86 systems and flexible storage, network, GPU, and DPU choices. Organizations can use purpose-built appliances, standard servers, external storage, disaggregated architectures, or hyper-converged designs according to workload economics and operational requirements. |
| Automate Day-0 Through Day-2 Operations | Provides web, CLI, API, provisioning, monitoring, and lifecycle workflows across infrastructure services, including scheduled and on-demand VM protection, controlled replication, storage health management, and repeatable operational recovery. |
| Scale Without Redesign | Supports single-node, clustered, disaggregated, hyper-converged, edge, and multi-site designs so compute, storage, networking, and acceleration resources can grow with demand while preserving a consistent operational model. |
STRATISYSTEM PLATFORM CAPABILITIES
StratiSYSTEM OS provides a common software-defined operating foundation for storage, virtualization, containers, Kubernetes, private cloud, AI, HPC, networking, security, observability, and lifecycle operations. Capabilities are activated according to the selected StratiSTOR, StratiSERV, or HyperSERV deployment role and the requirements of the workload.
| CAPABILITY | BENEFIT |
|---|---|
| Unified Infrastructure Operating Model | Provides common deployment, management, monitoring, security, policy, and lifecycle workflows across StratiSTOR storage, StratiSERV compute and virtualization, and HyperSERV hyper-converged infrastructure. The objective is operational consistency even when the underlying hardware, workload, or deployment topology changes. |
| Hardware-Independent Architecture | Runs on industry-standard Intel and AMD x86 platforms and supports flexible combinations of NVMe, SSD, HDD, external block storage, network adapters, GPUs, and DPUs. Infrastructure can be designed around workload requirements instead of a fixed proprietary appliance model. |
| Compute Express Link (CXL) Memory Support | Supports CXL-attached memory on compatible CPU, firmware, and CXL Type-3 device configurations through the Linux kernel memory subsystem. Configured CXL memory can extend host-addressable capacity when it is onlined as system RAM, while supported DAX configurations provide a separate device-access model. The StratiSYSTEM console (sdconsole) reports CXL presence, detected memory capacity, and operational status to distinguish available memory from detected but unconfigured or offline devices. Memory bandwidth, latency, and NUMA locality remain important workload-design considerations. |
| Composable Service Architecture | Allows storage, compute, virtualization, Kubernetes, private-cloud, AI, HPC, and networking services to be enabled according to the deployment role. Organizations can deploy dedicated storage or compute clusters, combine services with HyperSERV, or use mixed architectures as requirements evolve. |
| Web, CLI, REST API and Automation | Combines centralized web administration with command-line tools, REST APIs, scripting, and orchestration workflows for repeatable provisioning and day-2 operations. These interfaces support integration with broader automation, service-management, and infrastructure-as-code practices. |
| Role-Based Access Control and Multi-Tenancy | Supports role-aware administration, tenant isolation, policy-driven access, and centralized resource management for enterprise teams, managed-service providers, private-cloud operators, and shared infrastructure environments. |
| Centralized Observability | Provides health, utilization, performance, log, service-state, hardware, storage, network, VM, and cluster visibility from a common operational layer. This shortens troubleshooting paths and gives administrators a consistent view of infrastructure behavior. |
| Integrated VPP Accelerated Data Plane | Adds Vector Packet Processing (VPP) as a high-performance networking option for selected StratiSYSTEM data paths. Vectorized packet processing, multi-core scaling, and DPDK-capable I/O are designed to reduce per-packet overhead and improve throughput for demanding virtual-machine, container, storage, routing, overlay, and network-service workloads. |
| Advanced High-Speed Networking | Supports modern data-center networking with bonding and failover, VLAN, VXLAN, Geneve, GRE, SRv6, OVN, SR-IOV, RDMA/RoCE, InfiniBand-oriented fabrics, DPDK, VPP, QoS, traffic telemetry, and inter-datacenter connectivity. Designs can be optimized for low latency, high east-west bandwidth, or workload isolation. |
| PQC Fabric Secure Site Connectivity | Provides an optional site-to-site secure networking service with kernel WireGuard data transport, hybrid post-quantum key establishment, authenticated peer enrollment, managed key rotation, and FRRouting/BGP policy integration. Firewall ownership follows the deployed host role. |
| Hardware Offload and NUMA-Aware Performance | Supports architectures that use NIC offload, SR-IOV, DPDK, VPP, GPUs, and DPUs. High-performance deployments can be aligned to CPU NUMA domains, PCIe topology, queue configuration, and accelerator locality to minimize avoidable host-side bottlenecks. |
| Kubernetes Infrastructure | Kubernetes is included with StratiSERV, allowing virtual machines and cloud-native workloads to share a common infrastructure platform while using integrated compute, networking, persistent storage, monitoring, and lifecycle services. |
| Red Hat OpenShift Deployment Support | Red Hat OpenShift can be deployed on top of StratiSYSTEM when an OpenShift application platform is required, using StratiSERV compute and StratiSTOR persistent data services as appropriate for the design. |
| OpenStack Private Cloud | Supports OpenStack deployments for environments requiring broader private-cloud services, multitenancy, self-service infrastructure, API-driven consumption, and cloud-style resource orchestration. |
| GPU and DPU Infrastructure | Supports NVIDIA and AMD GPU-enabled infrastructure, GPU passthrough, supported virtual GPU technologies, and compatible NVIDIA DOCA and BlueField acceleration. These capabilities can be combined with high-speed networking and data services for AI, analytics, HPC, and accelerated infrastructure workloads. |
| GPU Monitoring and Thermal Visibility | Provides centralized visibility into supported GPU inventory, utilization, health, and temperature so operators can identify hardware or thermal issues, understand accelerator consumption, and make better placement and capacity decisions. |
| AI and HPC Infrastructure | Combines GPU-enabled compute, VPP/DPDK networking, RDMA-capable fabrics, StratiSTOR data services, Lustre, Kubernetes, and virtualization to support AI training and inference, simulation, analytics, rendering, scientific computing, and other data-intensive workloads. |
| Self-Hosted AI Model Deployment and Inference | Deploys supported GGUF models using local CPU, NVIDIA CUDA, or AMD HIP backends, with independent model services and one shared multi-model Open WebUI. Optional RAG knowledge bases, web search, persistent memory, TLS, and application API access support private inference workflows; distributed CPU inference is available for supported multi-node designs. |
| Lustre Parallel Data Services | Supports Lustre for workloads that require parallel file access and sustained aggregate throughput, including AI pipelines, research, simulation, rendering, analytics, and scientific computing. |
| Integrated Data Protection | Combines storage replication, erasure coding, snapshots, VM image-level backup, multi-disk protection, granular guest-file recovery, and scheduled or on-demand VM replication. Recovery and destination-replacement actions use explicit eligibility and safety checks. |
| High Availability and Failure-Aware Design | Uses clustered control and service patterns, fault-domain-aware placement, automated restart/failover mechanisms, and resilient data services to help infrastructure continue operating through disk, service, network, and node failures according to the configured protection model. |
| Secure and Disconnected Operation | Supports controlled and air-gapped environments with offline repositories and local services, role-aware access, encryption where configured, Secure Boot, virtual TPM, and supported confidential-computing technologies. PQC Fabric can provide separately managed authenticated, encrypted site links where deployed. |
| Flexible Deployment Models | Supports single-node, multi-node, disaggregated, hyper-converged, edge/ROBO, private-cloud, federated, and multi-site deployments while preserving a consistent operating model as compute, storage, network, and acceleration resources expand. |
| Migration and Modernization | Supports staged modernization through VM import and conversion from compatible VMware, Hyper-V, and local disk-image sources, shared storage integration, and traditional VM operation alongside container and Kubernetes workloads. Source prerequisites and compatibility constraints apply. |
| Lifecycle and Expansion | Standardizes service activation, maintenance, software updates, capacity expansion, and hardware growth. The platform is designed so compute and storage resources can be added incrementally rather than forcing wholesale infrastructure replacement as requirements change. |
STRATISTOR
StratiSTOR is the software-defined data-services layer of StratiSYSTEM. It provides scale-out block, file, object, and high-performance data services across supported NVMe, SSD, HDD, and external block storage. StratiSTOR is designed for enterprise applications, VM storage, databases, backup, AI datasets, analytics, media, archive, private cloud, and HPC use cases that require flexible capacity, resilience, and high-throughput data access.
FEATURES AND BENEFITS
| FEATURE | BENEFIT |
|---|---|
| Hardware-Independent Software-Defined Storage | Runs across supported industry-standard server platforms and storage media, allowing capacity, performance, networking, and failure-domain design to be matched to the workload. This supports cost optimization, supply-chain flexibility, and gradual hardware refresh without dependence on a single appliance configuration. |
| Unified Block, File and Object Services | Provides a common software-defined storage foundation for block, file, object, and transfer-oriented services. Organizations can consolidate multiple data-access patterns while retaining service-specific configuration and protection policies. |
| Scale-Out Architecture | Expands capacity, aggregate throughput, and fault-domain resilience by adding nodes and devices. Scale-out growth allows the storage platform to evolve with application demand rather than requiring a fixed controller pair or forklift upgrade model. |
| Flexible Media and External Storage | Supports NVMe, SSD, HDD, external block devices, SAN-attached capacity, and multipath configurations. All-flash, hybrid, capacity-optimized, and disaggregated designs can be built from the same operating foundation. |
| Enhanced Storage Performance with SDCache | Uses memory and caching algorithms to accelerate selected I/O paths and reduce backend media pressure. SDCache can be applied where workload locality and latency requirements justify dedicated cache resources. |
| Compression and Workload-Aware Data Reduction | Provides transparent compression and workload-aware data-reduction options, including supported LZ4 and Zstandard workflows and deduplication for eligible datasets. Protection and performance requirements determine policy selection; active VM datastores are excluded from the dedicated deduplication workflow. |
| Replication and Erasure Coding | Provides policy-based data protection using replication and erasure coding so availability, capacity efficiency, and failure-domain resilience can be matched to application requirements. Different protection approaches can coexist where the service design requires them. |
| Checksum-Protected Data Structures | Uses checksums, background scrubbing, and deep-scrub validation to detect inconsistent or corrupted data. Where a trustworthy redundant copy exists, repair workflows can recover affected data; persistent or repeated errors require device and cluster-health investigation. |
| Autonomous Maintenance and Self-Healing | Monitors device, placement, and protection health and uses distributed recovery and repair workflows after supported failures. Operators can inspect scrub status, inconsistent placement groups, recovery progress, and remaining redundancy before resuming normal maintenance. |
| Snapshots and Point-in-Time Recovery | Provides snapshot-based protection for rapid recovery, testing, rollback, and operational workflows. Snapshot policies can complement backup and replication without requiring a full data copy for every recovery point. |
| Non-Disruptive Cluster Maintenance and Expansion | Supports rolling service maintenance, node growth, and capacity expansion while data services remain available according to the configured redundancy and service architecture. Maintenance planning still accounts for workload, recovery state, and remaining fault tolerance. |
| Broad Protocol Support | Supports SMB v3, NFS v4, iSCSI, NVMe-oF, S3, HTTPS, SFTP, web-oriented access, and other supported data services. Legacy access methods can be retained where required while new workloads use modern block, file, and object interfaces. |
| High-Performance SMB, NFS and S3 Services | Provides scalable file and object services for enterprise data, analytics, backups, and AI datasets. Supported SMB Multichannel and tuned asynchronous I/O paths can increase throughput when client, server, CPU, network, and storage resources are appropriately configured. |
| NVMe-oF and Low-Latency Block Access | Supports NVMe over Fabrics for workloads that require efficient remote access to NVMe-class storage, complementing traditional iSCSI and other block-storage access models. |
| Parallel Data Services with Lustre | Enables Lustre and related high-throughput storage designs for HPC, AI, research, simulation, analytics, rendering, and other workloads that benefit from parallel file access and sustained aggregate bandwidth. |
| NVIDIA GPUDirect Support | Supports direct high-speed data paths between compatible GPUs and storage to reduce unnecessary CPU involvement and memory copies for AI, machine-learning, analytics, and other data-intensive workloads. |
| VPP-Accelerated Storage Networking | Provides VPP as an accelerated packet-processing option for selected storage and service data paths. VPP can complement DPDK, RDMA, SR-IOV, and conventional kernel networking where packet rate, latency, or CPU efficiency is a limiting factor. |
| Advanced Storage Networking | Supports high-speed Ethernet and workload-specific fabrics with bonding/failover, VLAN, VXLAN, RDMA/RoCE, InfiniBand-oriented designs, SR-IOV, DPDK, VPP, traffic policy, telemetry, and multi-network isolation for client and backend traffic. |
| Metadata Offload | Allows metadata-intensive services to place metadata on dedicated high-performance media, improving namespace responsiveness and reducing contention between metadata and bulk data operations. |
| Quality of Service and Workload Isolation | Provides mechanisms for controlling bandwidth, IOPS, network policy, and service placement so shared infrastructure can protect latency-sensitive or business-critical workloads from uncontrolled contention. |
| At-Rest Encryption | Supports encryption of stored data where configured, helping organizations protect sensitive information and address security or compliance requirements if physical media is lost, removed, or repurposed. |
| Public-Cloud Storage Integration | Supports public-cloud storage integration for selected backup, migration, archive, and data-movement workflows, allowing cloud services to be incorporated without making the local infrastructure dependent on a public-cloud control plane. |
| Synchronous Multi-Site Protection | Supports qualified multi-site designs in which writes are durably committed across protected locations before completion, providing an RPO of zero when quorum, placement, capacity, and inter-site latency requirements are correctly implemented. |
| Multi-Site and Multi-Cluster Replication | Supports protected data movement between sites and clusters for disaster recovery, geographic distribution, migration, and service-provider designs. Architecture is selected according to required recovery point, recovery time, latency, and operational model. |
| Active/Active Protected-Site Access | Supports designs in which protected data services can be accessed from more than one site when the selected service, placement policy, quorum model, and network architecture are validated for that topology. |
| Failure-Domain-Aware Placement | Allows protection policies to account for node, chassis, rack, and site failure domains so data placement reflects the failures the infrastructure is intended to survive. |
| OpenStack Integration | Integrates StratiSTOR data services with OpenStack-based private-cloud environments, providing scalable persistent storage for cloud workloads while maintaining the same underlying storage operating model. |
| Single-Node Through Large Scale-Out Deployment | Supports compact single-node or small-cluster use cases and scales to large multi-node environments. Resiliency, quorum, and service availability increase with appropriately designed multi-node deployments. |
| Centralized Storage Observability | Presents storage health, capacity, per-device utilization, placement, scrubbing, repair, recovery, service state, and performance metrics through the StratiSYSTEM management layer for faster diagnosis and capacity planning. |
| Simple Cluster Expansion | Adds capacity and performance incrementally by introducing new devices or nodes, allowing growth to follow demand while preserving the existing storage namespace and operational model. |
RECOMMENDED NODE SPECIFICATIONS
The following values are general design guidance rather than fixed appliance requirements. Final sizing should account for protocol services, workload concurrency, protection policy, recovery behavior, latency targets, media type, network topology, and the performance characteristics of the selected hardware.
| COMPONENT | OPTIONS / GUIDANCE |
|---|---|
| CPU |
|
| MEMORY |
|
| NETWORK |
|
| STORAGE |
|
STRATISERV
StratiSERV is the compute, virtualization, and workload-orchestration layer of StratiSYSTEM. It is designed to operate enterprise virtual machines, containers, Kubernetes, VDI, databases, AI workloads, and cloud-native services with integrated high availability, migration, protection, networking, automation, and observability. StratiSERV can use StratiSTOR, HyperSERV data services, or compatible third-party storage according to the deployment architecture.
FEATURES AND BENEFITS
| FEATURE | BENEFIT |
|---|---|
| Enterprise Virtualization | Provides production virtualization, clustering, virtual hardware, lifecycle management, networking, storage integration, and high-availability services through the StratiSERV operating model. |
| Live Migration (vMotion-Equivalent) | Moves running virtual machines between compatible hosts for maintenance, workload balancing, and lifecycle operations without requiring the guest workload to be shut down. |
| Storage Live Migration (Storage vMotion-Equivalent) | Moves virtual-machine storage between supported storage locations while the workload remains operational, supporting storage maintenance, rebalancing, tier changes, and infrastructure modernization. |
| High Availability | Detects supported host or service failures and restarts protected virtual machines on healthy cluster members according to available capacity and configured policy, reducing downtime after infrastructure faults. |
| Policy-Based Workload Placement | Uses cluster resource visibility and placement policy to help distribute workloads across available compute resources while accounting for operational constraints such as availability, maintenance, and accelerator requirements. |
| VM Templates and Rapid Provisioning | Standardizes virtual-machine builds through reusable templates and repeatable provisioning workflows, reducing manual configuration, deployment time, and variation between systems. |
| VM Snapshots and Checkpoints | Provides point-in-time virtual-machine recovery points for operational rollback, testing, patching, backup, and recovery workflows. Snapshot use can be coordinated with longer-term image-level protection. |
| Integrated Image-Level Backup and Restore | Protects virtual machines with multi-disk-aware image backups and full, alternate-VM, or supported disk-level restore workflows. Backup scheduling, immediate job execution, job-state reporting, and restoration safeguards help standardize operational recovery. |
| Granular Guest File and Folder Recovery | Inspects protected VM disks and restores selected files or directories without restoring the entire guest. Supports applicable Linux and Windows filesystems, including LUKS- and BitLocker-protected volumes when valid unlock material and guest-filesystem support are available. |
| Scheduled VM Replication and DR Controls | Supports scheduled and on-demand virtual-machine replication to designated recovery hosts, protected-source eligibility checks, destination inventory validation, and controlled handling of existing destination VMs for recovery or testing. |
| Shared Cluster Filesystem | Allows multiple StratiSERV hosts to access shared block storage for clustered VM placement and mobility, providing a VMFS-type operational model without tying the platform to a proprietary storage appliance. |
| StratiSTOR and Third-Party Storage Integration | Uses StratiSTOR or HyperSERV data services for tightly integrated deployments while also supporting compatible third-party block and file storage, enabling migration and coexistence with existing enterprise storage. |
| Cross-Platform VM Migration and Protection | Provides supported import and conversion workflows from VMware, Hyper-V, and local QCOW2/RAW images, alongside replication and protection mechanisms for infrastructure modernization. Migration eligibility depends on guest state, source access, snapshot configuration, and virtual hardware compatibility. |
| Hardware Passthrough | Provides direct assignment of supported PCIe devices such as GPUs, NICs, and other accelerators to virtual machines when near-native device access or specialized hardware functionality is required. |
| Virtual GPU Support | Supports compatible NVIDIA vGPU and AMD GPU virtualization technologies on supported hardware, allowing physical accelerators to be shared across multiple virtual machines where the vendor hardware and licensing model permit it. |
| NVIDIA GPUDirect Support | Supports high-speed GPU data paths for compatible AI and data-intensive workloads, reducing unnecessary CPU involvement and memory copies when combined with supported storage and networking configurations. |
| AMD ROCm and MxGPU Support | Supports AMD ROCm for GPU-accelerated compute and compatible MxGPU/SR-IOV virtualization capabilities, giving organizations an additional accelerator option for supported workloads. |
| GPU Inventory, Health and Thermal Monitoring | Provides centralized visibility into supported GPU inventory, utilization, health, and temperature to help identify hardware or thermal conditions and improve accelerator placement and capacity planning. |
| NVIDIA DOCA and BlueField Integration | Supports compatible NVIDIA DOCA and BlueField DPU technologies for networking, security, and infrastructure offload in AI, cloud, and high-performance designs. |
| Integrated Kubernetes and Container Orchestration | Kubernetes is included with StratiSERV, allowing virtual machines and cloud-native applications to run on the same infrastructure platform while using common compute, networking, persistent storage, monitoring, and lifecycle services. |
| Red Hat OpenShift Support | Supports deployment of Red Hat OpenShift on top of StratiSYSTEM for organizations that require the OpenShift application platform while retaining StratiSERV compute and StratiSTOR data services underneath. |
| OpenStack Private Cloud Integration | Supports OpenStack-based private-cloud environments that require tenant isolation, self-service infrastructure, API-driven consumption, and orchestration across compute, network, and storage resources. |
| VPP-Accelerated Virtual Networking | Adds VPP as a high-performance packet-processing option for selected virtual-machine, container, routing, overlay, and service-network paths, complementing SR-IOV, DPDK, OVN, and the standard virtual networking stack. |
| Advanced Virtual and Physical Networking | Supports VLAN, VXLAN, Geneve, GRE, SRv6, OVN, SR-IOV, RDMA/RoCE, InfiniBand-oriented fabrics, DPDK, VPP, bonding/failover, traffic policy, telemetry, and high-speed Ethernet for dense east-west and north-south workload traffic. |
| VDI with Data-Loss-Prevention Controls | Supports secure virtual desktop deployments with centralized infrastructure, policy, and access controls that can reduce endpoint data exposure while delivering remote or local desktop services. |
| API-Driven Automation and Global Tasks | Combines web administration, REST APIs, CLI tools, and cluster-wide tasks for VM lifecycle, backup, replication, job execution, networking, storage integration, placement, and maintenance workflows. |
| Role-Based Access Control and Tenant Isolation | Provides role-aware administration and resource separation for enterprise teams, service providers, private clouds, and shared infrastructure environments. |
| Performance Trending and Capacity Analysis | Tracks workload and infrastructure utilization over time so administrators can identify resource pressure, understand growth, and plan compute, memory, storage, network, and accelerator capacity before contention becomes service-impacting. |
| Full-Stack Hardware Monitoring | Correlates platform and physical hardware state to expose server, disk, NIC, GPU, temperature, service, and cluster health information through a common operational view. |
| UEFI Secure Boot and Virtual TPM | Provides modern guest-security controls for operating systems and applications that require trusted boot, TPM-backed identity, key protection, or current Windows security requirements. |
| Confidential Computing Support | Supports hardware-backed confidential-computing capabilities such as Intel TDX and AMD SEV on compatible platforms, providing additional isolation for sensitive virtual-machine memory and execution. |
| Per-VM CPU Compatibility Mode | Allows compatible virtual machines to use a controlled CPU-feature baseline so live migration can span supported CPU generations without forcing the entire cluster to a single global compatibility mode. |
| Air-Gapped and Controlled Environment Support | Supports disconnected or tightly controlled deployments using offline software repositories and local infrastructure services, helping address sovereign, industrial, defense, research, and restricted-network requirements. |
| Tiered, Disaggregated and Hyper-Converged Architectures | Supports dedicated compute, external shared storage, StratiSTOR-backed clusters, and HyperSERV designs so compute and storage can scale together or independently according to workload economics and operational preference. |
| Single-Node, Dual-Node and Scale-Out Clusters | Supports compact deployments and scale-out clusters up to the supported StratiSERV cluster limit. Availability, maintenance flexibility, and failure tolerance increase as appropriate cluster resources and redundancy are added. |
| Stretch and Multi-Site Architectures | When paired with the required StratiSTOR or HyperSERV data services, StratiSERV can participate in multi-site designs that maintain workload availability across site boundaries according to the configured storage, quorum, and network architecture. |
| Simple Cluster Expansion | Adds compute nodes without requiring a platform redesign, allowing CPU, memory, network, and accelerator capacity to grow as workload demand increases. |
| Centralized Observability | Combines virtual-machine state, cluster health, performance metrics, logs, hardware telemetry, networking visibility, and service status to reduce troubleshooting time and provide a consistent day-2 operations experience. |
RECOMMENDED NODE SPECIFICATIONS
StratiSERV sizing is workload-driven. CPU, memory, storage, network, GPU, and failure-domain requirements should be modeled together, with reserve capacity maintained for maintenance and node-failure scenarios in clustered deployments.
| COMPONENT | OPTIONS / GUIDANCE |
|---|---|
| CPU |
|
| MEMORY |
|
| NETWORK |
|
| STORAGE |
|
HYPERSERV
HyperSERV combines StratiSERV compute and virtualization with StratiSTOR data services on the same infrastructure nodes. The result is a hyper-converged architecture with a common control, monitoring, protection, networking, and lifecycle model. HyperSERV can be used as a compact HCI platform or as part of a broader StratiSYSTEM environment that also includes dedicated compute, storage, GPU, edge, or multi-site resources.
FEATURES AND BENEFITS
| FEATURE | BENEFIT |
|---|---|
| Integrated Compute and Storage | Combines the complete StratiSERV compute and StratiSTOR data-service stacks on the same infrastructure nodes, delivering virtual machines, Kubernetes, block/file/object storage, networking, protection, and management through one operating model. |
| Single Infrastructure Lifecycle | Coordinates deployment, monitoring, maintenance, expansion, and software lifecycle across compute and storage services, reducing the operational separation normally created by independent hypervisor and storage platforms. |
| Independent Resource Growth Where Practical | Allows cluster resources to be expanded according to demand. Additional HyperSERV nodes grow both compute and storage, while the broader StratiSYSTEM architecture can also integrate dedicated StratiSERV or StratiSTOR resources when asymmetrical scaling is required. |
| Mixed Workload Consolidation | Supports enterprise VMs, VDI, databases, Kubernetes, AI pipelines, file/object services, and general infrastructure workloads on a shared platform while retaining workload-specific placement, protection, and performance controls. |
| Integrated High Availability | Combines virtual-machine restart/failover mechanisms with resilient StratiSTOR data services so compute and storage failures can be handled within the same cluster architecture. |
| VM Mobility and Maintenance | Supports live VM migration, storage migration, host maintenance, and workload redistribution so planned infrastructure changes can be performed with minimal application disruption. |
| Unified Data Protection | Combines VM snapshots, multi-disk image backup, selective guest-file restore, and VM replication with StratiSTOR replication, erasure coding, snapshots, and multi-site data protection, supporting coordinated recovery across compute and storage. |
| Kubernetes and Cloud-Native Services | Includes Kubernetes through StratiSERV while using local HyperSERV data services for persistent storage, enabling VMs and containers to share the same hardware and operational foundation. |
| AI and GPU-Enabled HCI | Combines compatible GPU acceleration, high-throughput local storage, and high-speed networks with self-hosted CPU/GPU model inference, shared multi-model access, and optional knowledge retrieval services. Distributed CPU inference can use supported multi-node storage and compute arrangements. |
| VPP, DPDK and High-Speed Networking | Uses the same advanced networking capabilities available across StratiSYSTEM, including VPP, DPDK, SR-IOV, RDMA/RoCE, OVN, overlay networking, and 100GbE-and-faster designs for storage and workload traffic. |
| Centralized Management and Automation | Provides web, CLI, API, and automation workflows for compute, storage, networking, protection, and cluster operations from a common management plane. |
| Observability Across the Full Stack | Correlates VM, storage, network, hardware, GPU, capacity, and cluster telemetry so operators can investigate performance or availability issues without moving between unrelated infrastructure tools. |
| Edge Through Data-Center Deployment | Supports compact HCI footprints as well as larger clustered deployments, making the same operational model usable for edge, ROBO, private cloud, enterprise data centers, service providers, and distributed infrastructure. |
| Multi-Site HCI | Supports protected multi-site HyperSERV architectures when quorum, data placement, network latency, and capacity are designed for the required recovery objectives, including qualified synchronous designs for RPO-zero workloads. |
RECOMMENDED NODE SPECIFICATIONS
HyperSERV nodes should be sized as combined compute-and-storage systems. The design must account for VM/container density, storage performance and recovery, network bandwidth, accelerator requirements, HA reserve, and PCIe/NUMA topology so that consolidation does not create a resource bottleneck inside the node.
| COMPONENT | OPTIONS / GUIDANCE |
|---|---|
| CPU |
|
| MEMORY |
|
| NETWORK |
|
| STORAGE |
|