Network Functions Virtualization and the Need for Unified, Open Source Software

It has been a couple years since I last posted. I am not quite sure that I lost my way, but I certainly lost my motivation. Frankly, nothing really interested me. It was all garbage, a distraction or some ruse. And I needed to find something that I could be passionate about again. Well, Network Functions Virtualization (aka NFV) is just such a thing.

On the surface of things, the NFV was formed to do something that others have already been doing, which is to take physical appliances and run them inside of virtual machines. But this is where we depart from the pack and I get to become a bit more controversial. NFV really stands to overhaul how we build our data centers.

Now, there are plenty of people who’ll try to contrast NFV with say SDN (Software Defined Networking), but they’re probably quite stupid. The two have very little, if anything, to do with one another. More likely, you will find people equating NFV to being almost entirely about cheaper hardware. And they’d be wrong as well, although a bit more accurate.

NFV was started by Operators. For those of you not sure what I’m referring to, I’m talking about those companies that you probably take for granted as they provide you those data, texting and talky capabilities over your phone. Talk about an unappreciated bunch. Think about each time you Skype over your cell phone, just to avoid consuming minutes or long distance. Now picture the poor blokes that don’t get a pay raise because you just robbed their company of revenue.

Okay, so maybe it’s not quite that bad. But it is bad in their industry as they compete with the likes of the Googles, Alibabas and Tencents of the world. They’re the wire that transports the hot services. And it’s the guys making the hot services that are actually making the money and getting the ridiculous valuations.

Now, I’m not going to try to right the world here. Instead, I’m just gonna tell you that I see an opportunity and I’m capitalizing on it. That opportunity is to take various open source software — best of breed — and to unify the management interfaces at all levels.

It might be natural to think this to be a rather trivial exercise and something that vendors or even open source maintainers could do themselves, but I don’t believe that to be the case. Consider open source maintainers for a moment. They really don’t generally have the time or the motivation to do this. And what standard will they use, considering there really is no standard currently? It’d be more likely for a vendor to do this. But the challenge there is interoperability with other vendors. You see, vendors have a need to maintain some level of competitive edge. And while I would suggest that integration, testing, and building the best “whole” product is a great way of doing that, they will no doubt face the temptation to differentiate themselves at the management API, as well.

So, in my warped madness, I have decided to do what I have done for years now and to fork various projects. But rather than just take them and incorporate them into an appliance of my own, I’ve decided to make them available via a new portal I’m launching at Yúnifai.com over the next few weeks.

You might wonder how I came up with such a strange name. Well, it’s easy. It’s getting increasingly more difficult to get good, easy domains with a “.com” extension. So you have to be somewhat clever about it. Also, “Yún” (or 云) is the Chinese pronunciation for “Cloud”. I work for a Chinese company with over 150,000 personnel and find it amazing that no one else has drawn the connection to this. But at the end of the day, when we discuss unifying our various software and hardware components, aren’t we primarily doing it for the cloud? Well, that was my thought at least.

In addition to unifying (or maybe I should start saying Yunifaing) the management interfaces, we need to optimize the architecture and start getting rid of some of the excess abstraction layers and copies that go on in a virtualized world. NFV, no offense to Microsoft, is very likely going to be built upon Linux as the core operating system. If we’re talking about Linux as the Host OS and Linux as the guest OS, do we really need to replicate so much? I don’t think so. Instead, we need to strive for a collection of open source components that can work together in a highly optimal fashion. And that will require vision, a strong will and much tequila. I do not envision this to be an easy road ahead. And I’m sure many will consider this far too ambitious. And I can live with that. I’ve been told so many worse things over the years.

There are too many reasons why the communications and IT industries really need the “next” model for open source development. Think about what Linus does and leads inside of the Linux kernel. Sure, we’re talking about some brilliant software designers doing some fantastic work. But for all of that to work so well is largely the product of one highly anal individual and his drive towards perfection. Many fault him for this. I applaud him. This has to occur at the application and hypervisor level in order for NFV to really make much sense. Now, if you think I might be somewhat right on that point, then please tell me how it can be done more efficiently. I am always open to suggestions.

I am hoping that this renewed energy won’t die off any time soon.

Be back with more soon!

Mike

Leave a Reply

Your email address will not be published. Required fields are marked *