Enterprises have long had a choice of vendors in the networking space, but for most the dominant players have been Cisco and Juniper. The former is undergoing a reorg, and the latter is being acquired (subject to approvals) by HPE. While I’m not hearing enterprises express worry about this (see my blog on this), there is surely something afoot in network equipment, and that means there are both risks and opportunities in play. For some network vendors, the opportunity is clear, and Extreme Networks is one. Their announcement of their Platform ONE is surely a shot across the bow of the two networking giants.
Extreme has, for decades, battled for market share against giants like Cisco and Juniper. In this fight, as is often the case with a battle against giants, it’s been willing to deploy unconventional weapons. It introduced the notion of cloud-based management, and even a form of AI, well before it was prominent in the positioning of others, and even launched a network digital twin in 2022, as a means of gaining a systemic understanding of complex network infrastructure and its relationships to users and services. All of this seemed to be aimed at creating differentiation in a market where pushing packets is pretty much a matter of ones and zeros. Management and security seemed to be a good place to differentiate, and they still seem that way today.
Platform ONE is, as the name suggests, a broad tool. Its target is both network management and security, which hits a lot of enterprise hot buttons. The dual mission may be helpful to Extreme, because while security is a major enterprise priority with secure funding, there’s continuing skepticism about “platform” tools in the security space. It’s already got enough layers, say enterprises, and for most, Extreme isn’t a current layer provider but it wants to change that with a combination of cloud composability and AI.
AI is an almost-universal add-on to tools and systems these days, but it appears to me that Platform One is designed around AI rather than having AI plugged into it. In their presentation to analysts, Extreme AI Expert is the glowing core of the concept, in fact, the “One Ring” without the sinister Tolkien context. The principle is that networks are networks, and operationalizing and securing them as separate technologies or vendors is sub-optimal. Ops is best thought of as systemic, crossing management and security, LAN and WAN, virtual and physical. The more you know, the better it is.
The platform is then a cloud-hosted SaaS application, framed around an AI Expert core that in turn surrounds Extreme’s management/feature layer of tools, already hosted in the cloud. All Extreme product data is collected, and ecosystem partner data is likewise integrated. Other vendor equipment can be linked in via APIs, but it seems this would likely be the responsibility of channel partners or users to accomplish, at least at this point.
The user interface is both hierarchical and role-based, meaning that since Extreme’s sales conduit is largely based on resellers/integrators, the channel partner has a super-view of its customers, which then have their own set of role-based views within their own domain. Orchestration of the agent elements, governance and interface/data security, and platform service features are all integral to the Platform ONE core.
AI is integrated with the GUI features at all these roles/levels, and as noted it’s a core element of the platform and not a chatbot add-on. Three modes of AI operations are supported; conversational, interactive, and autonomous. In conversational mode, the AI element responds to user questions, much like a chatbot. This mode seems similar to a polled management framework; look when you want and see what you need. In interactive mode, the AI element will present conditions, like an event-driven system, and make suggestions, and the user can ask questions and implement recommendations. In autonomous mode, the AI element will actually take control and respond to things.
In terms of roles, Extreme offers two distinct classes of “users” (as opposed to channel partners offering integrated services). One class is called “users” and the other “buyers”, which may be a bit confusing, but reflects a distinction between those who operate the network and those who procure it. Things like budget planning and license and contract management fall into the latter category, while the former focus is the traditional operations elements. The progression of Learn, Plan, Deliver, and Fix is explicit in the design, with both the user and buyer classes having involvement in each step.
The goal of Platform ONE, in a functional sense, seems tied to workflows as a binder between infrastructure elements, network services, and user experience. Extreme has a longstanding interest in and support for virtual networking, and while the use of a virtual network is not mandatory with Platform ONE, I think it would enhance its capabilities by providing an explicit connectivity framework that can integrate the network environment.
Speaking of integration, one major question Platform ONE raises is how it’s adopted. Obviously, Extreme users can expect to leap into it and achieve real benefits. What about taking on Cisco or Juniper, though? The big money in networking these days is in the data center. The more cloud-centric an organization is, the less chance there is that there’s substantial network incumbency and an associated equipment transformation to deal with at the time of sale, but of course the less network there is, the less money there is in winning the customer in the first place. I think that real Extreme success has to come from actually displacing Cisco or Juniper, not from having the user shift away from the data center.
Would Platform ONE be enough, functionally, to justify a rip-and-replace? Probably not, unless either the competing gear was already old or there was a major change in network requirements that would justify replacing gear. Would it be enough to justify taking some or all of any planned network refresh? Yes, for many users, providing that it could deliver value to users during what Extreme would surely hope would be a network transformation in Extreme’s direction. That means pulling non-Extreme gear into the tent or depending on that “major change” to justify the refresh. The former approach is obviously safer and, if successful, more profitable. It’s also more appropriate for a channel-dependent vendor like Extreme.
Channel partners want leads more than anything else. They want their vendors to do the heavy lifting at the marketing and strategy level, generating excitement and prospects to call on. Even partners who have the skill and visibility to build their own leads will usually rely on vendors to pave the way in positioning. Extreme’s Platform ONE has the potential to generate excitement and leads, but the question of how it’s introduced into an account that’s not already using Extreme gear is important if Extreme is to take on, for example, Juniper as its customers navigate the HPE acquisition. Extreme CEO Ed Meyercord told CRN “Business doesn’t continue as normal when there’s a fundamental change like that. … And then that becomes a great opportunity to consider an alternative.” Like Extreme, obviously, and Platform ONE, with an appropriate Juniper bridge, could be just that. We’ll have to wait to see if that appropriate Juniper bridge, and the one for Cisco, is built. If it is, then both Juniper/HPE and Cisco might have to start looking over their shoulders.
