As we talk with telecom service providers about their plans for deploying Network Functions Virtualization (NFV), another topic invariably comes up. Depending on who you listen to, the Internet of Things (IoT) represents either a drag on network resources that leads to unrecoverable costs or else an opportunity for revenue growth through innovative new services. Or maybe both.
This post will highlight some of the implications of IoT for service providers, outlining potential upsides and downsides. This is hot industry topic and we’d welcome your thoughts on how service providers can best exploit this opportunity.
Frequently-quoted forecasts for IoT devices active by 2020 range from 25 billion (Gartner) to 75 billion (Morgan Stanley), with Cisco taking the middle ground at 50 billion. Regardless of which number is most defensible, there’s no doubt that the number of devices connected to service provider networks will be exponentially greater in five years than it is today.
So what’s the implication for the network itself of all these IoT devices that will be coming online?
First of all, it’s important to understand that the majority of “IoT-enabled” devices actually have no need for an Internet connection at all, though of course they may each be generating data which potentially requires additional network capacity. Sensors used in applications like home security, industrial control, connected cars and health monitoring systems typically connect locally with peer devices and/or with gateways. Mostly, it’s only the control systems which require an Internet connection back to the cloud (though some connected car applications are notable exceptions). This reduces the number of devices connecting to service provider networks by at least an order of magnitude, though we’re still talking about billions of new connections by 2020.
Once it reaches a service provider network, IOT traffic will present some unique challenges for the network infrastructure. Compared to regular Internet traffic, IoT will support so many different use cases that service providers will need to cater for a range of use cases spanning chatty, low-bandwidth connections all the way through to high-bandwidth persistent connections. The flexibility of NFV is key to providing the flexibility to address these diverse networking requirements.
Industry experts disagree widely on the expected bandwidth impact of IoT. On the conservative side, some are projecting that the overall amount of IoT traffic will be negligible compared to today’s Internet bandwidth, which is dominated by video and growing exponentially. Others take a more aggressive position, pointing out that once IoT becomes pervasive new use cases will emerge and traffic will explode.
Rather than raw bandwidth, the biggest network-related issues for IoT traffic are security, privacy, availability and latency.
Security and privacy will likely receive the most attention, at least in the media. As IoT devices become more and more prevalent in our lives, their usefulness for applications like healthcare, energy and home monitoring will demand their awareness of increasingly personal information, or at least information that easily be profiled back to you once it’s analyzed as “big data’ in the cloud. Service providers will need to implement sophisticated network security systems that meet the expectations of both consumer and enterprise customers.
Network availability will be critical for many IoT applications. If critical infrastructure systems like industrial control, emergency, healthcare and security are going to be sending time-critical traffic over a service provider network, they need to have a very high degree of confidence that the network will be up and all the networking services will be working. Enterprise-class availability (three-nines or 99.9% up uptime) won’t be adequate: these services will demand the six-nines reliability (99.9999% uptime) that today is only achieved by telco-grade networks. Again, there will be a wide diversity of requirements for network availability, reliability and resiliency: some IoT applications will tolerate packet loss while others will demand maximum fault tolerance.
The latency (and jitter) of service provider networks will have a strong impact on the usefulness of many projected IoT applications. If you’re driving a connected car with the expectation that traffic lights or sensors are going to react to your presence, a one second delay at 70 miles an hour means that you’ve travelled 100 feet. A lot can go wrong over that time and distance. Quality-of-Service (QoS) segmentation will allow service providers to position (and price) different service levels for different use cases.
All these issues point to the need for highly-reliable, low-latency, secure infrastructure platforms in service provider networks optimized for IoT traffic.
So how can the network be more than a “dumb pipe” and add real value to the enterprises consuming all this data that originates from IoT devices? One way is by intelligently determining what data is upstreamed and when that happens. Sophisticated gateways will segment their IoT traffic between information that needs to be transmitted to the cloud in real-time and data that can be stored and then transmitted when low-cost bandwidth is available, for offline analysis. This allows intelligent gateways to avoid constantly pumping massive amount of data “northbound”, overwhelming both the network and the storage infrastructure. It also enables them to provide different QoS for different payloads.
Software Defined Networking (SDN) technologies will use network Quality-of-Service (QoS) to manage traffic flows based on an awareness of the application or service, as well as the type of traffic. This awareness would come from the use of Deep Packet Inspection (DPI) and content inspection within an SDN framework. Essentially, SDN manages path capacity while NFV provides dynamic scaling of processing and networking resources.
One of the revenue-generating opportunities that service providers are discussing is the use of analytics to monitor and predict customer behavior, while assessing the state of the network. This can lead to automating the upselling and pricing of network services, such as temporary bandwidth expansion or ultra-low latency connections.
Finally, several service providers have talked to us about the value of “contextual knowledge” about their customers. Between their network infrastructure and their OSS/BSS systems (Operations Support / Billing Support), they capture continuous information about their customers’ location (if mobile), their characteristics of their network traffic and even the applications that they run. They have a unique business opportunity to turn this knowledge into value-added services.
What do you think? Is the explosion in IoT devices an opportunity or a threat for service providers? We’d be delighted to hear your thoughts.