Mobile networks are learning how to be webscale

Nokia Siemens Networks today laid out a new architecture for mobile networks that brings concepts such as automation and elasticity from webscale and cloud computing to mobile broadband. Much like the systems administrators trying to keep sites like Facebook or Amazon Web Services up have adapted their infrastructure of tens of thousands of nodes to be managed by a few people, network engineers at carriers are facing the same challenge.

What the webscale mobile network looks like

Billions of endpoints, from mobile phones to sensors, add complexity, while robust demand for mobile connectivity means more base stations in more places. As the scale of network equipment and demands increase, mobile operators are trying to do more with the same number of employees, or even fewer, so they can save on operational costs. Given how fast revenue from voice is declining this makes sense.

So on Tuesday NSN followed Alcatel-Lucent (s alu) in creating a network designed to scale and adapt to changing conditions and devices called Liquid Net (Alcatel-Lucent calls its gear LightRadio). From the NSN release:

“Capacity in today’s conventional networks is typically frozen in separate places; at individual base station sites, in parts of the core network that manage voice and data services, or in the optical and IP transport networks,” said Marc Rouanne, head of Network Systems at Nokia Siemens Networks. “Each is a potential bottleneck to someone getting the broadband service they want at a particular moment. Fluctuating, unpredictable demand in one part of the network means huge chunks of capacity can be left idle elsewhere, making poor use of existing investments. For example, as much as 50 percent of a conventional core network’s capacity can be dormant. Instead, Liquid Net unleashes frozen network capacity into a reservoir of resources that can flow to fulfill unpredictable demand, wherever and whenever people use broadband.”

This sounds pretty similar to the rationale for virtualization and now cloud computing, and the similarities aren’t an accident. As I wrote in January:

But multiple networks and more base stations, as well as more demand, are forcing operators to undergo a shift similar to what the data center saw as the demand for computing began to overwhelm the profits and abilities of systems administrators to handle it. For example, when it took one person to manage 10 servers, owning 500 was an investment, but now with corporations owning tens of thousands, such a ratio would constrain demand. So places that required a lot of computing adapted and came up with new architectures and software that helped become the redundant, autonomic and cloud-based computing centers familiar today.

Can telcos scale without open source?

However, as mobile networks attempt to scale, I wonder if they are at a disadvantage compared to folks building out webscale sites and compute clouds. The telecommunications world is full of very expensive, very proprietary gear and the “solutions” aiming to make mobile networks scalable are no different. NSN’s product works with a radio designed for it, and with layers of proprietary software. LightRadio is no different. And while there are startups and standards emerging around creating more network flexibility and scalability, the big gear makers still call the shots.

In the cloud world, there is much more emphasis on open source software and creating standards that work across platforms. This creates lower-cost infrastructure, but it’s also a constant source of technical innovation that the service provider world doesn’t really experience or even seem to demand. Sure, there’s VMware, (s vmw) but there’s also Citrix,(s ctx) Microsoft, (s msft) and now the OpenStack efforts creating rivals to an all-VMware cloud (of course, you can get that if you want).

Telecommunications providers are experimenting with open technologies such as OpenFlow as a way to cram more functionality in commodity boxes, but at this stage it’s just an experiment. I’m curious how far they might take it and embrace the more open ethos.

Forget the big guys, what does this mean for startups?

Just because the big guys are selling the gear for radio networks, base stations and the core network software, there is still room for companies trying to develop technology for preventing interference, monitor these larger networks and manage them, aggregate data on Wi-Fi networks to offer more capacity, and many other elements that will come into place to deliver enough broadband without breaking the bank for operators. Already companies such as Intucell, Sandvine, WeFi and others are delivering products that help service providers create flexible and heterogeneous networks that can adapt to applications in real-time. But the real avalanche would come if service providers embraced and more open and interoperable ecosystem.