Cisco leans on partners, blueprints for AI infrastructure growth

Cisco is taking a collaborative approach to helping enterprise customers build AI infrastructures.

At its recent partner summit, Cisco talked up a variety of new programs and partnerships aimed at helping enterprises get their core infrastructure ready for AI workloads and applications.

“While AI is driving a lot of changes in technology, we believe that it should not require a wholesale rethink of customer data center operations,” said Todd Brannon, senior director, cloud infrastructure marketing, with Cisco’s cloud infrastructure and software group.

As AI projects move from science projects in an organization’s backroom to mission-critical applications, enterprise infrastructure and operations teams are being challenged because they are dealing with new workloads running on familiar infrastructure but with new requirements, Brannon said.

“The idea is that we want to help our customers deploy and manage AI workloads efficiently, find that right mix of acceleration, and not over provision or leave stranded resources or create new islands of operations,” added Sean McGee, cloud & data center technology strategist with Cisco.

One of the ways Cisco intends to help customers is by offering a suite of validated designs that can easily be deployed as enterprise AI needs evolve. 

The company recently announced four new Cisco Validated Designs for AI blueprints from Red Hat, Nvidia, OpenAI, and Cloudera to focus on virtualized and containerized environments as well as converged and hyperconverged infrastructure options. Cisco already had validated AI models on its menu from AMD, Intel, Nutanix, Flashstack and Flexpod.

The validated designs allow customers to use these models and fine tune what they want to do for their business, McGee said.

Cisco is building Ansible-based automation playbooks on top of these models that customers can use with Cisco’s Intersight cloud-based management and orchestration system to automatically inject their own data into the models and build out repositories that can be used in their infrastructure, including at the edge of the network and in the data center, McGee said.

Cisco’s Intersight package manages a variety of systems from Kubernetes containers to applications, servers, and hyperconverged environments from a single location.

“Utilizing Intersight and our systems stack, customers can deploy and manage AI-validated workloads,” Brannon said. “The message is that we don’t want our customers and partners having to completely rethink the operation side, even though they’re having to rethink some things on the GPU provisioning side for AI, for example,” Brannon said.

In addition, as Cisco gets feedback from its customers on AI-specific features or additional validated designs, it will augment Intersight with new features, Brannon said.

Also, over time these models will evolve as more data is used to tune them, and customers can easily adjust them to fit the needs of their enterprise infrastructure, McGee said. “Our partners, too, can utilize these models to significantly expand their services. [They can] really give them a head start and relieve a lot of the engineering expense and time that they need to put these services together for customers.”

Cisco recently unveiled Data Center Networking Blueprint for AI/ML Applications that defines how organizations can use existing data center Ethernet networks to support AI workloads now.

A core component of the data center AI blueprint is Cisco’s Nexus 9000 data center switches, which support up to 25.6Tbps of bandwidth per ASIC and “have the hardware and software capabilities available today to provide the right latency, congestion management mechanisms, and telemetry to meet the requirements of AI/ML applications,” Cisco stated. “Coupled with tools such as Cisco Nexus Dashboard Insights for visibility and Nexus Dashboard Fabric Controller for automation, Cisco Nexus 9000 switches become ideal platforms to build a high-performance AI/ML network fabric.”

Cisco has also published scripts so customers can automate specific settings across the network to set up this network fabric and simplify configurations, Cisco stated.