Linux's NFV crew: Operators keen to ditch clunky networks, be 'cool' like, er, Facebook

So someone in charge of a virtual network project

... says network operators are wanting to switch to virtual networks. Hmm, old Heather there wouldn't be a teensy bit biased would she?

"want to ditch their clunky networks to compete for "cooler" consumer services"

Buzzword bullshit. Next...

"She believes that on the operator side, the existential threat is from web companies. They can't compete at the same speed because their proprietary networks are slow,"

Err, sweetie - Facebook and Google are web services companies, they sit on top of the network layer and ultimately use the same lines as everyone else, they don't provide it. What they do internally on their LANs is irrelevant. I think you need to go have a little lie down.

again completely irrelevant

But whenever I see and hear anything about a hipster Renaissance fair I always think of that scene in The Cable Guy where they go to the medieval food place and chuckle. Come back here, so that I may brain thee!

Good luck with that

OPNFV are fighting the good fight, but really, technology has nothing, zero, zip, nada to do with it. Web culture succeeds because "good enough" is enough to get started with and iterate on later.

To take an example. Let's say you want to virtualize, your BGP routing function. "Hmm, I don't like paying Cisco and Juniper for big boxes. What to do?"

Facebook: I see some open source stuff. It's a bit poor, but maybe that is a good place to start. OK, let's start with a basic router, with 10% of the function, and go from there. I'll buy some helper tech, hire some people, etc. Telco: I see some stuff. But oh dear, it can't do the 1000 features I'm used to. And I can't offer 100 features, it has to be the full 1000. Result: quelle surprise, the only people who you can go to are...Juniper and Cisco. Oh dear. Running IOS/JunOS in a VM is not exactly NFV is it? And what happened to the savings that consultant told me about? Where's my NoSQL? (sorry, couldn't resist, see https://www.youtube.com/watch?v=b2F-DItXtZs)

Telcos are, as ever, desperate to be out of their core business -- shifting bits. They very much want to move "up the stack" to where the money is seen as greater. Although given that telcos are near the top of most nations league table of stock, you've got to wonder at the wisdom of this ambition.

NFV is seen as a way to both reduce costs and to move up the stack.

As far as reducing costs, telcos have made a dreadful error with the design of 3G: installing bespoke high-touch devices. These bring in some revenue for telcos ("header enhancement"), protect the fragile wireless network from abuses such as DoS and scanning, and provide network functions such as NAT. But when you look at the 3G design you'll see large numbers of these boxes, each purchased at a huge margin (because you pay more for box labelled "3G system firewall" whatever is inside it).

So the attraction of NFV for telcos is moving all these expensive 3G systems into VMs, and then managing those VMs economically and well.

Now for the revenue side. The main focus here is enterprises. Want your employees to have secure access to interior resources without the fuss of a VPN? Then same for cloud-delivered services? But also content providers: want a CDN deep in the wireless network so your content is streamed from next to the wireless head-end?

I'm pessimistic on the revenue side. If only because telcos have a habit of stuffing that up. So many telcos tried movie delivery, but then added so many T&Cs on that and at such a margin that it was difficult and expensive to use. So Netflix rolled right over the top of them. So quickly that telcos couldn't organise a response until it was all over. I imagine there will be a few worthwhile applications, but nothing like the revenue streams some people hope for. The tidying of telco 3G networks will be worth it however: mobile networks have a bad habit of going extensively titsup for hours due to failures in high-touch equipment.