Q&A: What Are All These AI Tools Going to Do to Higher Ed IT Infrastructure?

EDTECH: AI is being positioned as versatile technology that can help with almost everything a higher education institution does. Where do you see AI being most useful?

NISPEL: There is certainly AI for networking and AI for data and how you engage with it. I'm mainly focused on how we can use AI to change the way users interact with our systems that we provide. We're building products for and with AI to make people's lives easier in terms of network operations, how they configure applications, how they interact with security and how they interact with enterprise data overall.

EDTECH: Speaking of data, institutions are going to need increased storage and computing to handle the amount of information input and output through AI tools. How can they best prepare?

NISPEL: What you need to do in network infrastructure to optimize for AI workloads is going to fall on the data center, obviously, but that is right now more on the hyperscale side. If you look at higher education, certainly, you will see more and more computer science departments jumping onto the GPU bandwagon and building up capacity. So, you need to ensure that those workloads can hold up in terms of latency and throughput storage.

We're also seeing requirements, even at the network edge, that are changing for AI. And especially in higher education, there's a lot of Internet of Things and smart building infrastructure that is being built that will generate data. Then the question becomes, where do you process the data? More and more edge compute workloads are required for AI inference, but also for distributed learning, so you need to make sure that you take that into account. We're seeing all of those moving at the same time.

EDTECH: This may be hard to answer, but is there any way to anticipate just how much more storage and computing power institutions will need in the coming years?

NISPEL: That is a difficult question. We recommend CIOs take into account that, as the classroom gets smarter, adoption of AI from a student perspective gets bigger, and that new tools such as augmented and virtual reality are also being used. There's a lot of change; the density is changing a lot, and the best strategy to plan ahead and protect yourself from shortage is to rethink your Wi-Fi access infrastructure.

We have recently, over the past two years, seen new capacity become available in Wi-Fi 6E and Wi-Fi 7. The 6-gigahertz band that’s now available is basically doubling the access capacity for Wi-Fi, which then makes Wi-Fi more interesting for critical workloads. CIOs should definitely prepare on the access layer side for 6GHz and rethink the wireless access infrastructure for the higher density that's going to happen.

Capacity needs to be adopted at the edge, though, and I think that's critically important. And as we have distributed learning, don't just look at the Wi-Fi access, look at how remote access and security are being covered as well in that context.

On the data center side, obviously, those workloads are very different when it comes to learning, computer science or experimenting. There, I think, most CIOs need to rethink the data center infrastructure as well. Network fabrics such as the one we offer at Extreme Networks allow you to scale in hundred-gig increments in the data center. This becomes really important because institutions can scale the data center very effectively and reuse that same technology on campus to converge student networks, research networks, facilities management — just about everything — in one architecture.

EDTECH: Are there other recommendations you have for future proofing these networks and data centers so they will be easier to scale whenever the next emerging tech is announced?

NISPEL: In terms of scaling for Wi-Fi specifically, the 6GHz band has different propagation characteristics, so you need to redesign your network for 6GHz. But then, when you move to Wi-Fi 6E or 7 or 8, it will not matter because this is the big change that everybody needs to be aware of. Rethinking and redesigning the network for that frequency band, and the coverage that you want to achieve and the experience that you want to deliver, is going to be critical.

If you have a higher ed deployment with as many as 15,000 access points, however, you are going to need an upgrade cycle. What we recommend is to have a specific refresh plan. Don't let it lapse, so to speak. As you build out, it's not a one-time thing. You're continuously building out, and you're continuously refreshing. So, don't build up too much technical debt by waiting too long because you think you want to write off your deployment — suddenly, you won’t be able to move anymore, and it will take you five years to upgrade the infrastructure. Specifically on Wi-Fi, it's driven by the nature of the technology. Every couple of years there's a new Wi-Fi standard, then new chips coming out, so that is inevitable. Don't try to fight it. Embrace it, and make it part of the plan.

READ MORE: Universities might consider these five design issues when upgrading to Wi-Fi 6E.

On the flip side, we've also seen a trend in network operations and automation toward the cloud, which gives you a certain level of abstraction where the underlying infrastructure can change but the operating model doesn't change. And you always have the scale and capacity in the cloud to make those changes along the way. That is certainly something that should be considered. We have seen a push to the cloud in all verticals, including in higher ed, for management and operations specifically.

On the switching side, the same strategy applies. But with our Fabric technology, you can more easily expand capacity without redesigning all the time. I think that's a clear advantage for us, and we think we have a pretty good architecture in place that gives way more flexibility.