I recently spoke with Vish Nandlall, CTO for Ericsson North America, about telecommunications, WebRTC, and the Web of Things.
Ian: Vish, what do you make of news that in the US, the FCC wants to move the telephone system to IP?
Vish: The battle between IP and TDM (Time-division multiplexing) has left telecom with a lot of scar tissue. Architectural choices are informed by the job to be done. My professional life was concerned with scale and robustness of networks —tight integration between a few services: voice, messaging and access. Interoperability between different networks was at the service layer. As the basis of competition shifts to scale and flexibility of applications, the transition to IP is inevitable. IP decouples services from networks —it is a new backplane between and within operators, that allows scale adoption of new services without the need for service layer interoperability between different networks. IP is mature, reliable, and performs well enough for this industry transition from mobile telephony to mobile compute. I embrace it.
IJ: Who do you think will most feel the impact?
Vish: In North America, we’ve already been making the transition to IP. The Long Term Evolution (LTE) mobile network is IP-based, for example. However, globally we are still in the early days of the transition. As the app marketplace continues to grow, those in North America will have an advantage. On the wireline side, it’s a completely different story. The install based of TDM access is still quite large. It’s no mystery that there’s been a huge migration from fixed line voice to wireless. This has allowed operators to replace infrastructure with IP-based infrastructure. The increases in revenues have come from Ethernet circuits not TDM.
IJ: What are the business needs driving Web adoption?
Vish: The business model disruption of the Web is the the death of distance. This drove social and retail change by sparing consumers the “transaction cost” of attending a high school reunion, or visiting a store. Physical location no longer confers advantage.
Vish: This model is reversed in the post-PC era. The mobile Web delivers offline context to online behavior. With a smartphone, I can gather information online and compare prices as I browse products in the aisles. The new business driver is to frame all my resources at the point of activity. In an enterprise context, this manifests as a worker being able to access dispatches from workforce management tools, linked to robust communications to access skills within the field services organization.
Vish: This requires rethinking our 19th century model of communications. For instance, unlocking the value of voice by innovating around all the components of a voice session. WebRTC can create unique opportunities by taking voice into new contexts such as Web telephony, anonymous calling, permission based calls, group calling, voice messaging, dynamic call routing, do-it-yourself IVRs, call referral tracking, and more.
IJ: Is this the app story essentially?
Vish: I tend to interpret the app story through the lens of my background in hardware. When I look at communications, I wonder “what is the Moore’s law” that will spur the transition from one platform to another, and create the conditions for acceleration? In my view, WebRTC provides the foundation for tens of thousands more communications applications. WebRTC enables the proliferation of endpoints that can be integrated into many communications applications. I want to be able to use different identities, different signaling mechanisms, different security architectures (more for business-to-business, less for ordinary social conversation). I think that the interesting thing about WebRTC is that it prescribes none of those and allows all of them.
IJ: How much of this is moving up the stack (and disrupting the existing the power structure)?
Vish: The control point for telecommunications has been the phone number. This enabled lock-in to the platform and made it hard to bust out of that architecture. If identity becomes fungible — I can use my email address or Facebook id or telecom id interchangeably — that creates an interesting tension and new opportunities, but also the risk of gaps and fragmentation. I think the ecosystem will help to fill those gaps (as jQuery has). We’ll continue to see tension between frameworks, libraries, and potential fragmentation.
IJ: Do you think that the platform being open helps to mitigate those risks?
Vish: The more open, easy to use abstraction layers you have, the more network effect you can get (as well as niche opportunities). Let’s look at the numbers. There are about half a million developers that can be addressed in orthodox, tightly coupled, telecom protocols, but about 5 million developers who understand modern-day methods based on loosely coupled RESTful and JSON interfaces. When you get a step function like that —an order of magnitude shift in the number of developers who have the requisite skill and talent— you end up with outsize network effects, capable of meeting the demand created by the proliferation of devices and the Internet of Things.
IJ: Are there skills that you think Web developers need to acquire to function more effectively in a telco environment?
Vish: Developers look at a network as an infinite pool of bandwidth, almost as an abstraction of the network. They throw packets into the network and count on them getting to the other end. But networks are messy — you have to deal with jitter, latency, and other channel quality issues. If you try to run something as simple as web sockets on a wireless network you may experience a lot of drops due to session timeouts. So developers need to consider questions like how to hide the complexity when the session drops. There are many other issues to deal with as well, from battery consumption to processor architecture. These physical world issues must be considered in context. Events need to be discovered, understood, and interpreted, and you may not have answers baked into standards.
Vish: As the balance of power shifts to endpoints, developers will need to know how to interact with them. The telecommunications industry has many opportunities and a role to play. It’s exciting to bridge the worlds of call processing in IMS, HTTP, Session Initiation Protocol (SIP), and mobile access points hung off traditional circuit-switching. These changes are causing confusion for business models, but I think we are creating a rich set of capabilities, especially with WebRTC to scaffold them. Now you can do things in 2 weeks that used to take 18 months. I think this agility will lead to more experimentation with business models.
IJ: Any obstacles to getting there?
Vish: WebRTC is important as the technology enabler, but you also need software abstractions, a community of developers, the means for monetization, social structures to publicize apps, and the means for content creation and distribution. We have focused on software and developers but not looked closely yet at monetization, which is important to creating viable developer platforms. Ericsson has been working with Dialogic (telepresence), Tropo (app servers in telco environments), our own IMS platforms, and operators who are interested in running these types of experiments. Until we have these ecosystems and the platform capabilities, it will be somewhat fragmented.
IJ: One strength of the Web is the ability to scale to a large number of devices of all types. How is Ericsson seeing the approaching Web of Things?
Vish: As a connectivity provider, we’re very interested. But you need a different way to manage a network when you go from 100 million to 1 billion connections in a single network (or from 5 to 50 billion globally). From the business model perspective, you have to be more efficient to lower the cost per connection. You can’t connect a refrigerator to the network for $50 per connection. What does business look like at $1 per connection? We’ll need to move to both licensed and unlicensed spectrum. We’ll need to accommodate both high-power and low-power networks. And we will need to integrate them in a way that delivers capacity at reasonable cost, through automation and lowering the number of administrators per CPU. To create business potential we’ll need the right software abstractions, greater ability in how we deploy services, and cloud computing. We’ll need to approach our IT systems more from an enterprise perspective. This is all inevitable. As we put these enablers into the business context, we’ll see the acceleration.
IJ: What do you mean by acceleration?
Vish: Connecting sensors, homes, cars, production lines and so on entails a certain amount of economic and technological friction. I can gain efficiencies by communications-enabling every business process. Suppose I’m having a conversation with a supplier. I want to be able to log the call and find it later in my CRM system. Automating that, for example, would be a simple improvement that would create big efficiency gains. I want to be able to embed voice and make it searchable. Or suppose I’m in a seminar where the vocabulary is challenging. I’d like to transcode the voice in real-time, extract key terms, and link to explanations to accelerate learning. If I look at industries that have not benefited from big productivity gains — the service industry as a whole, or health care — I think it’s because they have not yet availed themselves of these technologies. There is great potential when we embed telecommunications in business processes.
IJ: These sound like tough integration challenges. What approaches might help us meet those challenges?
Vish: We’re starting to crack the code on integration. Loose integration of RESTful APIs reduces the integration time scale. This progress comes from the ecosystem of tools and loose integration through APIs.
Vish: When I look at workflow APIs used in data centers, they are becoming more instrumented and make it easier to visualize faults. Remote patient care is another example. You can go from examination to prescription by integrating a wireless video link, health care records, telecommunications, and a collaboration platform. You can put systems like this together in 2 months instead of 2 years. We’ll see even more complex integration down the line as we get better at creating abstractions.
Ian: Thank you, Vish!