Infortrend Enterprise Cloud Family

Infortrend Enterprise Cloud (IEC) is a private cloud platform that integrates compute, storage, and Kubernetes management into a single solution. It offers cloud capabilities such as scalability and reliability, enabling enterprises to run both cloud-native and traditional applications. IEC deployment is recommended to start with three identical compute nodes, each featuring high-performance CPUs, GPUs, and NVMe SSD drives, making it suitable for AI and other high-performance applications. The platform includes node protection, supports seamless scale-out to enhance performance, and delivers high-performance storage for demanding data needs. With its built-in Application Marketplace, IEC enables enterprises to deploy various applications quickly and effortlessly.

The IEC compute nodes KS 5000U are available in three model types. They feature configurable options for GPUs and U.2 NVMe SSDs, allowing enterprises to select the model that best suits their application needs. All KS models have a 2U form factor and support 25GbE and 100GbE network interfaces.

GPU-Intensive Model – KS 5008U

KS 5008U supports up to 4 high-performance NVIDIA GPUs and 8 NVMe SSDs per appliance, with scale-out capabilities of up to 80 GPUs per cluster. Users can choose between dual Intel® Xeon® Scalable or AMD EPYC™ 9004 Series CPUs, which provides adaptability to various performance and workload demands.

The system is designed for optimized GPU resource utilization, ensuring multiple applications can efficiently share GPU resources and allowing the system to maximize GPU performance.

With high GPU density and efficient resource allocation, KS 5008U is well-suited for AI training, generative AI, and 3D rendering. It is also well-suited for industries increasingly relying on AI or GPU, such as healthcare, retail, manufacturing, and Media & Entertainment.

Supports up to 4 NVIDIA GPUs per Unit

Scalable to 80 GPUs per Cluster

Optimized GPU Utilization

General-Purpose Model – KS 5016U

KS 5016U supports up to 2 high-performance NVIDIA GPUs and 16 NVMe SSDs per appliance, with scale-out capabilities reaching 40 GPUs and 320 NVMe SSDs per cluster. This balanced configuration, combined with dual Intel® Xeon® Scalable CPUs, delivers efficient processing and minimizes bottlenecks.

Designed for workloads that require GPUs with higher memory and processing power, KS 5016U is well-suited for AI applications such as deep learning inference, chatbots, and AI services, ensuring smooth handling of large datasets and real-time responses. Also, this model is a perfect fit for HPC and AI-driven manufacturing processes that demand intensive computational resources.

General AI Server for Diverse AI Applications

Supports High-Performance GPUs

Scalable for Future Growth

Storage-Intensive Model – KS 5024U

KS 5024U supports up to 24 NVMe SSDs per appliance, with scale-out capabilities reaching 480 NVMe SSDs per cluster. Powered by dual Intel® Xeon® Scalable CPUs, it delivers high-speed storage performance with large capacity, making it ideal for storage-intensive workloads. These include general IT applications such as email servers, databases, and CRM applications, as well as manufacturing environments where factory equipment continuously generates operational data.

High-Speed Storage with NVMe SSDs

Scalable to 480 NVMe SSDs per Cluster

Scalable for Future Growth

High Performance

GPU Computing: Each KS node supports up to 4 Nvidia GPU cards, which makes it ideal for AI, deep learning, and high-performance computing (HPC) applications.

Exceptional Computing Power: With each KS node featuring 64-core computing power, the entire cluster can provide up to 1280-core computing power.

High-Speed Storage: Equipped with U.2 NVMe SSDs, the KS node provides storage delivering 1M IOPS, 30GB/s read, and 25GB/s write storage throughput performance for running applications.

Scalability

Expand compute nodes: Users can start with one KS node or a three-node cluster (recommended). The cluster can be expanded non-disruptively by adding one node at a time as compute requirements grow, ensuring resource optimization for applications.

Expand storage capacity: KS supports various storage expansion options, allowing connection to EonStor GS unified storage, EonStor CS scale-out NAS, or EonStor GSx parallel file storage.

Flexible Data Protection by Application

KS supports multi-node data replication (Replica x2 or x3) and allows replication to be configured individually for each application. Critical applications, such as manufacturing systems, databases, and CRM applications, can be configured with tailored redundancy policies (Replica x2 or x3) to ensure optimal data protection. For less critical applications, such as non-production workloads, replication may not be necessary, allowing you to allocate resources where they are most needed. This approach provides greater flexibility in data protection management.

Backup and Restore

For greater flexibility, KS offers a backup of containers and VMs. By backing up all application data and settings, KS ensures consistency for containerized applications, helping enterprises maintain seamless operations.

Enterprises can streamline their data protection strategies with remote backup to cloud or remote storage. To improve backup efficiency, KS offers a variety of features, including scheduled backups, incremental backups, and a comprehensive monitoring dashboard to oversee backup status and ensure the process runs smoothly.

High Availability

KS is designed with automatic node failure detection and failover functions. The system can automatically detect node failures and quickly transfer applications to healthy nodes. With application protection, KS ensures that even in the event of a physical node failure, the application services continue to run without any downtime.

Faulty node can be completely removed from and rejoined to the cluster system after repair, without requiring system downtime. This greatly enhances availability and ensures the continuity of enterprise applications.

Automatic Scaling

KS can dynamically scale cloud-native applications across nodes according to real-time node load conditions. It achieves this by continuously monitoring the hardware load of computing nodes, including metrics such as CPU, GPU, memory, storage space, and network usage.

When computing demands for an application increase, the system reallocates resources by automatically scaling the application to underutilized nodes. Conversely, when the demand decreases, KS automatically downscales resources to optimize efficiency. This dynamic adjustment ensures consistent performance and efficient resource utilization.

Additionally, KS resolves a key auto-scaling challenge in the open-source Kubernetes version. For stateful applications that rely on databases, scaling is often problematic because data between nodes is not synchronized. Our solution allows stateful applications to scale seamlessly by storing data separately in external storage, ensuring consistent access and preventing data mismatches, even as the application scales.

Optimized GPU Resource Utilization

GPU sharing for resource efficiency: This allows multiple applications to share the same GPU resources, enabling applications to compute simultaneously while dynamically adjusting GPU resources based on each application's needs.

GPU passthrough to boost performance: This allows VMs to directly access physical GPU resources, achieving performance nearly equivalent to running directly on the host. This feature addresses the typical limitations of VMs, where GPU performance is constrained by hardware virtualization. It significantly benefits GPU-intensive applications such as 3D rendering, machine learning, and HPC.

Application Market

KS offers an Application Market with applications across more than 10 categories, including manufacturing, IT services, AI/machine learning, big data, DevOps tools, e-commerce, and business applications, addressing the diverse needs of customers. Enterprises can also create custom applications on KS and share them across different departments, effectively reducing redundant work. Infortrend offers KS for running applications but excludes implementation services, requiring customers to build or deploy their own applications.

User-Friendly Management

KS offers a user-friendly solution for efficient management with EonOne and EonKube graphical user interfaces.

EonOne provides comprehensive hardware monitoring features to ensure smooth system operation and optimal performance. Additionally, it includes a simplified management interface, making it easy for IT staff to scale out computing nodes or expand external storage nodes.

EonKube is designed for application management, enabling enterprises to manage both containerized applications and virtual machines on a single platform. This facilitates a smooth transition to cloud-native architecture. The built-in Application Market provides applications across various fields, effectively addressing diverse business needs. With its intuitive GUI interface, users can deploy applications easily, eliminating the need for complex deployment files such as YAML. Additionally, EonKube comes with robust monitoring and security features, ensuring precise resource management and enhanced system performance.

Support for Cloud-Native and Traditional Applications

Run Different Application Types: KS provides a built-in Kubernetes platform for running cloud-native containerized applications. With a featured hypervisor, KS enables support for traditional applications, such as Microsoft applications, by running VMs. This allows enterprises to consolidate all workloads on a single platform.

Various OS Compatibility: KS supports various mainstream Windows and Linux versions, offering extensive operating system compatibility to meet diverse business and application needs.

Easy Deployment

KS features an intuitive installation wizard that guides users through the entire setup process, which can be completed in under 30 minutes. By streamlining the workflow and integrating hardware and network diagnostics, KS ensures a smooth and efficient deployment.