Select a tab
Cisco Reveals Nexus HyperFabric AI Clusters, A New Simplified Data Center Infrastructure Solution with NVIDIA for Generative AI
Hybrid cloud infrastructure for AI and the next decade with Cisco Unified Computing System (UCS) and WWT
Application growth along with intensive demands for AI and Machine Learning workloads has forced organizations to rethink infrastructure and operations. In a hybrid cloud world, IT must free itself from the burden of managing massive complexity in order to keep up. Cisco UCS X-Series Modular System with Intersight empowers organizations to simplify IT and harness the full the full power of AI to drive innovation and maintain a competitive edge in a data-centric landscape.
WWT experts and the AI Proving Ground in our Advanced Technology Center (ATC) support the latest in Cisco UCS and Intersight solutions.
This unique lab environment accelerates your ability to learn about, test, train and implement innovative AI solutions.
UCS C885A
The Cisco UCS C885A M8 Rack Server is a high-density GPU server designed to deliver massive, scalable performance for AI workloads. Built on NVIDIA HGX architecture (with options for NVIDIA H100/H200 Tensor Core GPUs) or AMD MI300X OAM GPUs, the C885A M8 accelerates compute-intensive tasks like large language model (LLM) training, fine-tuning, and large model inferencing, and Retrieval Augmented Generation (RAG). Get faster results from your AI initiatives, improve productivity and business efficiency, and modernize your data center infrastructure with this powerful and versatile solution.
- Unparalleled Performance for AI/ML: Equipped with 8 high-performance NVIDIA HGX H100/H200 Tensor Core GPUs or AMD MI300X OAM GPUs, the C885A M8 delivers massive, accelerated computational performance in a single server. High-speed GPU interconnects like NVIDIA NVLink or AMD Infinity Fabric mesh enable demanding AI model training jobs.
- Scalability for Demanding Workloads: Scale your AI workloads across a cluster of Cisco UCS C885A M8 servers to address deep learning, LLM training, model fine-tuning, large model inferencing, and RAG. Each server includes one NVIDIA ConnectX-7 NIC or NVIDIA BlueField-3 SuperNIC per GPU to scale AI model training across a cluster of dense GPU servers.
- Flexibility with Configurations: The C885A M8 offers fixed configurations optimized for intensive AI and HPC workloads. Choose from various configurations with different CPUs (4th or 5th Gen AMD EPYC), memory (up to 3TB DDR5), and storage options (up to 16x 2.5" U.2 NVMe SSDs) to meet your specific needs.
- Simplified Management with Cisco Intersight: Managed by Cisco Intersight, the cloud-delivered IT operations platform, the C885A M8 helps reduce TCO and increase business agility. Intersight provides intelligent automation, observability, and optimization for your AI infrastructure, allowing you to operate with consistency and control.
- High-Speed Networking: Features 400G East-West NICs (NVIDIA ConnectX-7 or BlueField-3 B3140H) for high-speed inter-chassis, inter-GPU connectivity to perform training at scale. North-South connectivity is provided via NVIDIA ConnectX-7 or BlueField-3 B3220.
In today's rapidly evolving landscape, AI is no longer a futuristic concept but a core business imperative. The Cisco UCS C885A M8 Rack Server provides the robust infrastructure you need to accelerate your AI initiatives, gain a competitive edge, and drive innovation. By providing massive compute power, simplified management, and flexible configurations, the C885A M8 empowers you to unlock the full potential of AI and transform your business.
UCS C845A
The Cisco UCS C845A M8 is a versatile, accelerated 4RU rack server designed to handle a multitude of AI workloads. Built on the NVIDIA MGX platform, this server offers exceptional scalability and flexibility, allowing you to drive performance for demanding AI applications, from large-scale model training and fine-tuning to high-performance inferencing. Its modular architecture and air-cooled design ensure it integrates seamlessly into your existing data center, providing a future-proof solution for your evolving AI needs.
- Scalable AI Performance: Configure with 2 to 8 NVIDIA H100 NVL, H200 NVL, or L40S GPUs to match your specific workload requirements. Easily scale up the number of GPUs as your AI demands increase. The UCS C845A M8 also supports AMD MI210 GPUs.
- Optimized for GenAI: The C845A M8 is designed to excel in generative AI applications, including large language model training, fine-tuning, and inferencing. The NVIDIA H200 NVL GPUs can significantly boost inference speed, and the server's architecture supports Retrieval-Augmented Generation (RAG) for enhanced AI capabilities.
- Consistent Management with Cisco Intersight: Unified management of your AI servers alongside your traditional infrastructure through Cisco Intersight. Intersight delivers foundational capabilities for server management, including claim, inventory management, health monitoring, power control, and basic monitoring. Post-FCS, Cisco Intersight will enhance UCS C845A M8 server management through a phased feature release, including advanced capabilities such as server profiles, real-time updates, OS and firmware management, and automated deployments.
- Flexible Configuration: Choose between high-frequency AMD 9575F or high-core-count AMD 9655 5th Gen EPYC processors. Configure with up to 3TB of DDR5 memory (32 x 96GB at FCS). The system also offers flexible networking options with NVIDIA ConnectX-7 or BlueField-3 NICs for both East-West and North-South traffic.
- Extended Lifecycle and Serviceability: The NVIDIA MGX architecture allows for the integration of next-generation GPUs without costly redesigns. Key components like DIMMs, CPUs, and other components can be replaced individually by CX Service Support, ensuring a longer and more serviceable lifecycle compared to traditional GPU servers.
In the era of AI, having the right infrastructure is critical. The Cisco UCS C845A M8 provides the performance, scalability, and manageability you need to accelerate your AI initiatives and gain a competitive edge. From research and development to enterprise applications, the C845A M8 empowers you to unlock the full potential of AI and drive innovation across your organization.
UCS X-Series Modular System
The Cisco UCS X-Series with Intersight is a modular system managed from the cloud or an on-premises appliance, designed to support both traditional enterprise and modern workloads in a hybrid cloud environment. Utilizing Intel® 4th and 5th Gen Xeon® Scalable Processors, and soon to include AMD processors, the UCS X-Series features a unified form factor, eliminating the need for multiple IT silos as application diversity and volume increase. This system enables organizations to scale infrastructure and adopt new technologies, such as increased GPU processing power, without standalone upgrades of compute and networking resources.
Recognized with the 2023 SEAL Sustainable Product Award, the UCS X-Series is purpose-built for a sustainable future. Cisco Intersight's AI-driven operations further enhance efficiency with predictive analytics, automating routine tasks and proactively detecting issues before they impact performance, thus optimizing the infrastructure for the complexities of AI-driven transformations.
The Cisco UCS X-Series 9508 Chassis, known for its exceptional modularity, supports a variety of GPU configurations to meet diverse computational needs. It can accommodate GPU blades such as the H100 and the L40S, or smaller mezzanine adapter GPUs like the T4, with each blade connecting to two smaller form factor GPUs. Larger GPUs require the X440P blade, complemented by the X-Fabric deployed in a dual fabric on the back of the chassis. The modular design allows for easy swapping of parts, akin to connecting Lego pieces. Additionally, the new UCSX-Direct, a FI-less solution with miniaturized FIs placed in an A/B pair at the back of the chassis, has been announced and is now orderable. This makes it an ideal solution for AI at the edge, as it retains all the benefits of the standard data center chassis without the need for FIs at the top of the rack. The ability to incorporate a wide range of GPU processing options enhances its suitability for edge computing, offering flexibility, robustness, and powerful AI capabilities in a compact design.
The Cisco UCS X210c M6 and X210c M7 servers are designed to be seamlessly managed through Intersight, whether deployed on-premises or in the cloud. Both servers offer robust performance and flexibility for a variety of workloads, but they differ in several key areas. The X210c M6 server is powered by Intel® 3rd Gen Xeon® Scalable Processors and supports up to 6TB of DDR4 memory, making it an excellent choice for general-purpose computing needs. In contrast, the X210c M7 server features the latest Intel® 4th and 5th Gen Xeon® Scalable Processors, offering enhanced performance, increased memory capacity of up to 12TB with DDR5, and advanced security features. The M7 server also provides superior support for AI and ML workloads with its advanced processing capabilities, higher core counts, and improved memory bandwidth. These enhancements enable more efficient handling of complex computations and large datasets, giving the M7 a significant advantage over the M6 in AI and ML applications. Both models benefit from Intersight's comprehensive management capabilities, which streamline operations, enhance security, and enable predictive analytics for proactive issue resolution, ensuring that organizations can maintain optimal performance and efficiency in their IT environments.
The Cisco UCS X410c M7 server, featuring a four-socket Intel® 4th and 5th Gen Xeon® Scalable Processors, stands out as the higher performance option. It offers superior processing capabilities, higher core counts, and increased memory bandwidth, making it ideal for intensive AI applications such as real-time data analytics, natural language processing, and complex neural network training. Upgrading current GPUs is effortless, allowing organizations to adapt their AI strategy by swapping out GPUs for lower or higher-end models as needed. This modularity ensures that adding or upgrading GPUs is hassle-free, enhancing the overall flexibility and scalability of the UCS X-Series system. This integration allows organizations to maximize their computational capabilities, efficiently handling complex tasks and large datasets within a unified infrastructure. For those not requiring high-end GPUs, the M6 and M7 X210c blade servers also offer a smaller mezzanine option, accommodating up to two smaller NVIDIA T4 GPUs, providing versatile solutions for varying computational needs.
The Cisco UCS X440p, also known as the Cisco X-Fabric, is a proprietary fabric designed to seamlessly integrate high-end GPUs with the X210c M6 and M7 servers within the X9508 chassis. This innovative fabric offers an easy-to-use, modular approach to connecting powerful GPUs, such as the NVIDIA H100 and L40S, to the servers. Inserting the X440p into the back of the chassis is straightforward, making it simple to enhance your system's GPU capabilities. The X-Fabric supports a variety of NVIDIA and Intel GPU configurations, enabling both the M6 and M7 servers to leverage advanced processing power for demanding workloads, including AI and ML applications. Upgrading current GPUs is effortless, allowing organizations to adapt their AI strategy by swapping out GPUs for lower or higher-end models as needed. This modularity ensures that adding or upgrading GPUs is hassle-free, enhancing the overall flexibility and scalability of the UCS X-Series system. This integration allows organizations to maximize their computational capabilities, efficiently handling complex tasks and large datasets within a unified infrastructure. For those not requiring high-end GPUs, the M6 and M7 X210c blade servers also offer a smaller mezzanine option, accommodating up to two smaller NVIDIA T4 GPUs, providing versatile solutions for varying computational needs.
UCS B-Series Blade Servers
Tap into better performance and operational efficiency for a wide range of physical and virtual workloads to include AI-optimized, scalable blade infrastructure for high-performance computing and analytics.
UCS C-Series Rack Servers
Tackle any workload challenge with a flexible combination of processor, memory, I/O, and internal storage resources, which provides versatility and AI-ready rack servers for demanding data workloads and machine learning.
UCS S-Series Storage Server
Enable scalable, high-capacity solutions for both general data center and AI-driven applications with this storage-optimized server.
Fabric Interconnect
Unifies computing, networking, and management resources into one cohesive system for UCS servers.
Management Platform: Cisco Intersight
Unified management across Cisco UCS and third-party storage, servers, and networks, also offering cloud-based AI analytics for management, optimizing data center resources and automating workflows.
Cisco+ Hybrid Cloud
Cisco+ Hybrid Cloud is a way to manage hybrid cloud with a simplified, consumption-based solution that delivers flexibility, predictability, and visibility.
Cisco Data Center certfications and awards
WWT employs 367 Cisco-certified engineers and holds more than 6,060 individual Cisco certifications. Notable certifications include many CCIE data center certifications, including:
- Cisco Gold Integrator
- Cisco Master Data Center and Hybrid Cloud Specialized
- Cisco Advanced Data Center Specialized
- Cisco Hybrid Cloud Computing Specialized
WWT was also recognized with three Cisco global partner awards in 2023, including Americas Partner of the Year and also the global Security & Collaboration Partner of the year.
Learn more about Server Infrastructure and Cisco UCS
AI Proving Ground
Cisco AI Toolkit Demonstration
WWT's Deployment of Codeium on Cisco UCS with NVIDIA L40 GPUs: A Success Story in Security and Productivity
How to Choose the Right IFM for Your Cisco UCS X9508 Chassis: 25G vs 100G
Why Organizations Should Transition to Cisco UCS X210C M7 Servers
Embracing the Future: Transitioning from Cisco UCS 6200 Series Fabric Interconnects
Understanding Cisco UCS Management with Cisco Intersight
Server Infrastructure