I have a very fond memory from when I was 10, when my father pulled out the newspaper and asked my younger brother and me about going to the movies. There was buzz about a film that had just opened—a story set in a galaxy far, far away. The rest is history. I’m an unabashed Star Wars nerd (just ask my two daughters), and when Cisco chose to use “hyper” to name two of its latest solution offerings, I was immediately reminded of the Millennium Falcon using its hyperdrive to make the jump to hyperspace.
With that “important” context set, I’ll tell you that Cisco held its signature North American customer and partner event in Las Vegas during the first week of June. To no one’s surprise, AI was a central theme and, as usual at the event, the launch payload was substantial. Three announcements stood out to me as especially noteworthy. The first two serve as the inspiration for my Star Wars–themed introduction, but the third is equally out of this world. Let’s dive into the details as I provide insights related to each.
Hypershield For Security
Cisco’s Hypershield offering was initially announced in April. Hypershield represents a clean-slate, ground-up redesigned security architecture aimed at protecting data centers in the context of the generative AI gold rush. New AI workloads are taxing existing infrastructure in terms of compute, networking and security. It’s the latter challenge that Cisco is addressing with Hypershield, and on the surface this represents a significant leap forward for the company in its journey to provide a deeper security portfolio.
Cisco highlights three capabilities that make Hypershield unique. First, an AI-native architecture is designed to deliver fast response and continuous protection. Second, it has the ability to scale to protect a wide expanse of enforcement points with a single unified policy construct. Third, it employs a fabric that is architected to incorporate incremental enforcement without the need to rewrite policies.
From my perspective, Hypershield represents an opportunity to raise Cisco above the AI hype cycle noise and deliver practical end-to-end protection that employs autonomous segmentation and self-qualifying upgrades. Next-generation AI workloads will eventually move to the network edge, given the recent introduction of AI PCs. This will undoubtedly increase the complexity for securing language models, data and the underlying infrastructure at scale; Hypershield is well positioned to support the distributed nature of an evolving technology stack.
Hypershield is in customer preview now, so it’s too early to call it a success. However, the announcement at Cisco Live to support AMD Pensando DPUs by the end of the year (something I analyzed in detail in a recent research paper) points to the company’s commitment to broadening its next-generation security architecture—as well as Hypershield’s specific potential. It seems to be paying off: At the event, Cisco executives highlighted overwhelming customer interest for Hypershield, as evidenced by a long waiting list for evaluation.
Nexus HyperFabric For Generative AI Workloads
Nexus HyperFabric is Cisco’s attempt to simplify the deployment and management of AI clusters, promising a single-click experience that couples Cisco switches with other vendors’ offerings to support generative AI applications. The partner offerings include Nvidia GPUs and DPUs, other makers’ DPUs, VAST data storage and the Nvidia AI Enterprise software stack. Nexus HyperFabric is also cloud-managed to operate at massive scale, and it’s designed to leverage tried-and-true Ethernet as well as Cisco’s depth in optics and custom silicon to deliver the performance needed for demanding AI workloads.
I like that the interconnect is Ethernet versus Nvidia InfiniBand, because Ethernet is a standard that has existed for over 50 years, and one that delivers a good balance of price and performance. If anything, it should get even better over time, given ongoing industrywide efforts to improve the Ethernet standard. What I also like about Nexus HyperFabic is that it leverages Cisco’s strengths in networking and marries it with Nvidia accelerated computing and software. In my mind, this truly reflects a partnership of equals, unlike other recent Nvidia partnership announcements such as the one with Dell Technologies—given Dell’s shallower networking depth relative to Cisco. Nexus HyperFabric is currently in customer preview, with general availability expected by the end of this year.
Digital Experience Assurance For Operational Efficiency
Digital experience monitoring is hardly a new concept, as evidenced by all the available solutions from not only Cisco but also Fortinet, Netskope, Zscaler, IBM (most recently via IBM Concert) and others. I’m confident that Cisco kept this point firmly in mind as it architected its new Digital Experience Assurance platform. DXA aims to further leverage the company’s acquisition of ThousandEyes, and I had an opportunity to be briefed about it by executives ahead of the announcement at Cisco Live. For me, the single most compelling aspect of DXA is its ability to provide visibility into both owned and unowned infrastructure and enable users to take informed action through network assurance.
Unowned infrastructure includes public cloud provider services and anything else that is not directly managed by enterprises. Often, it’s these elements that are difficult to monitor, which makes it hard to detect when something isn’t functioning properly that affects the quality of connectivity and SasS application experiences. DXA brings a view of owned and unowned infrastructure into a single management console, potentially radically simplifying network assurance and any remediation steps. It does so while incorporating several new features including ThousandEyes Cloud Insights, Traffic Insights, Endpoint Experience and a new Meraki Assurance Overview—all of which are AI-infused, and all of which could enrich visibility for owned and unowned infrastructure, supporting a global area network construct.
The new DXA features are currently in customer preview, with general availability expected by later this year and early next year. However, some capabilities are already available to customers that leverage ThousandEyes assurance data on the backend, including AI radio resource management for Catalyst wireless and capacity planning for Catalyst SD-WAN services, as well as device profiling with AI-based signatures for Cisco Identity Services Engine.
Wrapping Up
At Cisco Live U.S. 2024, chief executive Chuck Robbins spoke to AI’s ability to deliver limitless potential in a complex world. From my standpoint, generative AI could deliver a force multiplier effect greater than that of the cloud, given its potential disruptive impact for consumers and enterprises alike. Cisco clearly wants to cash in on this, and its Hypershield, Nexus HyperFabric and DXA solutions all point to enormous potential for the company to do so.
Proof will come in the form of flawless execution and customer acceptance, and it seems that the company is marshaling the necessary resources to make that a reality. As a famous philosopher once said (in the second installment of the Star Wars saga): “Do or do not. There is no try.” The same could be said about the eventual enterprise infrastructure winners in the AI era.
