As AI/ML initiatives develop to new heights, the community has change into a crucial enabler for innovation and enterprise development. At Cisco, we acknowledge that at the moment’s AI clusters can now not be confined inside a single information middle. As a substitute, a “scale-across” strategy is required to seamlessly lengthen AI workloads and optimize sources throughout a number of information middle places.
Energy constraints and resiliency necessities are inflicting hyperscalers, neoclouds, and enterprises to embrace distributed AI clusters that span campus and metro areas, all of which want safe, high-performing, high-capacity, and energy-efficient connectivity. That’s why we’re excited to introduce the Cisco 8223, our newest addition to the Cisco 8000 fastened sequence, optimized for large-scale disaggregated materials inside and throughout information facilities, enabling clients to scale AI infrastructure with unmatched effectivity and management.
Assembly the calls for of distributed AI
The Cisco 8223 is a power-optimized fastened router, making it very best for environments with restricted energy. With 51.2 Tbps of capability and excessive programmability, it provides organizations most flexibility to adapt to evolving networking wants. Supporting each Octal Small Kind-Issue Pluggable (OSFP) and Quad Small Kind-Issue Pluggable Double Density (QSFP-DD) optical type components, the 8223 is purpose-built to attach geographically dispersed AI clusters. In an period the place AI mannequin coaching and inferencing require seamless interconnectivity and large information throughput, the 8223 stands out with its capacity to offer constant, high-bandwidth, low-latency communication throughout a number of information facilities.
Listed here are a number of the 8223’s key improvements.
Powered by Cisco Silicon One P200
On the coronary heart of the 8223 is the groundbreaking Cisco Silicon One P200 51.2 Tbps, deep-buffer routing chip. This chip delivers unequalled throughput whereas sustaining the bottom energy consumption, empowering organizations to scale sustainably from small to large site visitors spine networks.
Deep-buffer and high-radix structure
The 8223’s deep-buffer design offers ample reminiscence to briefly retailer packets throughout congestion or site visitors bursts, a necessary characteristic for AI networks the place inter-GPU communication can create unpredictable, high-volume information flows. Mixed with its high-radix structure, the 8223 permits extra units to attach immediately, decreasing latency, saving rack house, and additional decreasing energy consumption. The result’s a flatter, extra environment friendly community topology supporting high-bandwidth, low-latency communication that’s crucial for AI workloads.
Versatile working system assist
The Cisco 8223 helps a wide range of community working programs (NOSs), together with open-source choices corresponding to SONiC, with deliberate assist for Cisco IOS XR. Moreover, Cisco Nexus 9000 Sequence Switches, powered by Silicon One P200, are deliberate to supply assist for NX-OS. This strategy meets a variety of necessities and permits clients to tailor their networks to satisfy their particular wants.
Built-in safety
With inline MACsec for data-in-transit safety, the Cisco 8223 allows seamless networking to assist safe, latency-sensitive AI workloads throughout distributed information middle environments.
Superior optics
Help for coherent optics as much as 800GE in OSFP and QSFP type components is offered, enabling high-speed, long-distance connectivity for high-performance networking.
Compact, scalable type issue
The Cisco 8223 seamlessly integrates into current infrastructure or serves as the inspiration for brand new deployments. Its compact 3 RU fastened type issue design gives operational simplicity and ease of scaling, whether or not enhancing present capabilities or constructing out new distributed AI clusters.
Product advantages embrace:
- Accelerated time-to-value: A deep integration with a number of NOSs, together with open-source software program ecosystems, delivers agility and operational flexibility.
- Enhanced sustainability: Business-low energy consumption helps inexperienced IT initiatives and long-term value financial savings.
- Efficiency optimization: Deep buffers forestall packet loss, and inline MACsec ensures your information is protected finish to finish.
Shaping the way forward for AI Networking
The Cisco 8223 ushers in a brand new period of scalable, safe, and environment friendly networking for distributed AI clusters. As your group appears to leverage the facility of AI throughout areas and information facilities, the 8223 ensures your community can maintain tempo by delivering the efficiency, flexibility, and peace of thoughts that at the moment’s digital leaders demand.
Go to Cisco throughout OCP International Summit at Sales space #B22, the place these fastened programs can be on show.
Take a look at the Cisco 8000 Sequence Routers touchdown web page
Further Assets: