After my blog on a model-driven service lifecycle management technique, I got a bunch of emails from operator and vendor contacts (and from some who’d never contacted me). Part of the interest was driven by a Light Reading article that noted my skepticism about the way that ETSI NFV was being implemented. Most was driven by interest in the specifics of the model that I think would work, and it’s that group that I’m addressing here.
My ExperiaSphere approach to NFV has been extensively documented HERE in a series of tutorials, and I’ve now added another tutorial to the list, this one cutting horizontally across the service-lifecycle-stage approach taken by the earlier tutorials. I’ll be adding it to the ExperiaSphere tutorial list, but in the meantime the new presentation can be found HERE.
My goal in this extra tutorial is to link the event-centric evolution in public cloud services I’ve blogged about, with the needs of network service lifecycle automation. This was a principle of ExperiaSphere from the very first (which some of you may remember was in 2007), but the specific features of Amazon, Microsoft, and Google can now be used to explain things in a modern-relevant way.
The key point here is the same one I’ve made in the earlier blog I referenced, and that I made to Carol Wilson in the interview for the Light Reading piece. We have to do NFV using the most modern technology available, which means “event-centric” cloud and data-model-driven portability of features. We know that now because of the direction that all the major cloud providers are taking.
We’ve built our concept of networking on the notion that a network connects addresses, which represent real and at least somewhat persistent things. We’re entering an age where the concept of addresses, persistence, or even “things” is restrictive. In the cloud, there’s no reason why features can’t migrate around to find work, rather than the other way around. There’s no value to a specific resource, only to resources as a collection. Users are transient things, and so are the services they consume. This is the future, both of the cloud and of networking.
All of this was pretty clear a decade ago, and so were the challenges to promoting the vision. I was a part of a group called the “IPsphere Forum” or IPSF, and it was probably the first initiative that saw services as things you composed on demand. The founding force behind it was a vendor, which prompted the vendor’s arch-rival to try to torpedo the whole notion. Operators jumped on it and worked hard to bring it to fruition, but they were defeated in part by regulations that forbid their “collusion” in operator-dominated standards work. And above all of this was the fact that at the time socializing a concept this broad and different was difficult because most in the industry had never even thought about it.
They think about it now. We’re now seeing the future of networking differently, and in no small (and sad) part because we’re really seeing it in the cloud and not in the network. In networking, everyone has focused on protecting their turf as market changes and competition threaten their profits. There was no massive upside to connection services, no big benefit cloud to fight for. Cloud provider competition, rather than trying to protect the status quo, is trying to advance the cloud faster into new areas. Regulations don’t impair cloud providers. Most of all, for the cloud we’re sitting on a trillion-dollar upside.
That upside could have gone to network providers, both services and equipment. Networking was, and still is, the broadest of all tech industries in geography and “touch”. It can tolerate low ROIs, and has plenty of cash to invest. As an industry, networking has squandered a host of advantages it had in defining the fusion of network and computing that we call “the cloud”. As an industry, networking has even largely squandered itself, because its future is now out of its own hands.
ExperiaSphere is my attempt to frame the future in at least a straightforward way. Maybe it’s not something everyone understands, but I think every network and IT and cloud professional would understand it. I want to emphasize here that I’m not “selling” ExperiaSphere, or in fact selling anything related to it. The material on the website is all open and public, available to anyone without attribution or fees. I’m selling an idea, a vision. Better yet for the bargain-conscious, I’m giving it away in these tutorials and my blogs.
This blog is posted on LinkedIn, and anyone who has a view on the issues I’ve raised can comment and question there. There’s an ExperiaSphere Google+ community too and I’d be happy to take part in discussions there as well. Do you share the vision I’ve cited, or do you have reasoned objections? Let’s hear your views either way.