What trends are you seeing in enterprise network architecture?

We are seeing a huge shift to cloud 

You can fit a decentralized architecture in a distributed model. But if you look at the trends and you follow integration companies, you're hearing about things like distributed data architectures. There’s this notion that you're going to have some single store of data, but very few people ever get there and stay there long because of tech debt and things that have happened in the past. If you look at integration and network architectures, they're talking about fabric and mesh. I think that we’ll end up in a world that presumes distribution; with distribution you can collapse things. The distributed end has more flexibility than the others do in it, which is why I think we’ll see that. You can attempt to unify everything, but the world moves on before you ever finish. Those are the trends I'm seeing — it’s unclear whether people are actually out there executing in this way yet. They might be doing it for certain advanced applications, but we've been talking about this for 20 years and we haven't gotten there yet.

In terms of the paradigm shift, I know we have been slowly, steadily concentrating into public clouds. If you think about decentralization or edge devices in the modern paradigm, it’s the modern version of legacy data centers. The data centers are still there and a lot of companies are holding onto them right now. They have not fully adopted the cloud, so they want to figure out if there is a shortcut: If you have all the infrastructure and investments sitting in the data center, is there a way they can be reprovisioned into an edge cloud kind of a paradigm, without throwing it all away?

I'm seeing a definite shift now that multi-cloud and hybrid cloud have come to be. People are realizing that edge can benefit them in two ways. The first benefit is proximity — to be closer to your customer, react faster and maybe even be proactive. The other benefit is latency, which is becoming an issue for a lot of people, particularly when they're doing analytics. Companies that  are going through transformation or shifting to industry 4.0 are finding similar problems.

This iteration of the web is one for all, and the next iteration is all for one. Because in this iteration, there’s the one-for-all model represented by Twitter, Facebook, or some other kind of collaborative environment that serves as one platform for all users. I dictate the platform and the ownership of the data, which is all mine; I’m free and clear to use it because I'm giving you the privilege of using my network. That's the Facebook and Twitter model, which is why they resell data. Now that people are much more aware that they’re being tricked into contributing their expertise or knowledge to a large community, nobody should have the right to resell that data without attribution to its original authors.

In the next iteration, we're now looking between decentralized and distributed. If you're in the crypto or NFT world, it tends to be very distributed. If you're in the enterprise world, it's tending towards decentralization. As for their data centers, you can take one data center and make it 200 data centers in a heartbeat by using colocation or other data center providers. You cannot do that with instances of a cloud. That's unheard of. I'm hearing a little death knell around the big cloud service providers (CSPs); they’re going towards a plateau and a decline because more and more applications are becoming necessary for corporations to use that require proximity and latency to be considered in their design points.