Managing the Internet of Things: The network operator’s challenge
The Internet of Things (IoT) is exploding - and its traffic is totally unlike anything that network operators have encountered. Here’s how savvy carriers can respond to this challenge, and use standards to turn the IoT from a management and security headache into a business opportunity to offer new agile services.
Network operators and service providers are gearing up for the IoT - and the IoT mixes revenue opportunities with technology challenges.
Consumers are going crazy for IoT devices, from wearables like the Apple Watch and Fitbit health band, to Internet-connected thermostats, to the Amazon Echo cloud music player to connected vehicles. Businesses, too, love the IoT, using IP-based technologies for inventory control, smart signage and medical devices.
It’s a huge market, estimated to reach US$1.7 trillion by 2020 in direct products and services, with a $263 billion support market there will be billions of connected “things” in the IoT, with estimates ranging from 13-25 billion by 2020.
While many technology-driven conversations about the IoT focus on the end points (like fitness bands) and the back-end applications (like social networks that track and analyse exercise), the truth is that the IoT depends entirely upon safe, secure, highly available connectivity.
For the end point, that’s everything from home and coffee shop WiFi access points to 4G cellular data services to business internet. On the back end, service providers manage IoT from the device to data centres or the cloud, tied together with MEF CE 2.0, MPLS and fibre optical networks.
IoT presents opportunities for the telecommunications companies that provide that connectivity, and for industry vendors that provide hardware, software and services to the carriers.
It presents challenges as well, because IoT traffic has different characteristics than traditional network traffic, explains Marie Fiala Timlin, CENX director of marketing, which offers lifecycle service orchestration solutions for Software-Defined Networks.
“Internet of Things traffic is characterised by high-volume signalling and low-bandwidth data traffic. There are a lot more events happening in the network, which leads to a lot more data being collected. And when I'm talking about data, it's not necessarily the consumer application or subscriber-type data - it’s network events happening,” Timlin says.
She continues, “IoT exacerbates the big data network problem and creates a need for service providers to even more efficiently manage their network, because of all these events that are happening in the system, for example, whether it's for troubleshooting or for ensuring high quality of service.
"That really drives the need for lifecycle service orchestration, because here you've got a big data problem, and you're trying to apply all the cloud computing technologies that have already been used to solve big data analytics problems in the business world.”
Planning for traffic impact
“Because the IoT is such a fast-growing market,” says Dr. Hongwen Zhang, CEO and co-founder of security services provider Wedge-Networks, “when carriers try to do infrastructure deployment, and they have to do the capacity planning, they are looking at a three-to-five year planning cycle. Maybe longer.
"How do you handle IoT with a fast-growing future, with limited capital? One thing that can provide the solution is Software Defined Networks and Network Functions Virtualisation. Spend your money there, and don’t over-invest in firewalls and switches.”
Timlin adds, “Operators have to track network events for each network device and the connection in order to keep the pipe up and running. They also have to aggregate subscriber-level events, such as which application is being used.
"The best of both worlds is actually marrying those two types of data so you can intelligently determine when and where you need to augment network capacity. That's where NFV comes into play, because then you can do that augmentation much more flexibly and cost-effectively.”
That means analytics - lots of analytics - about network traffic, says Angus Robertson, insightsoftware.com vice president of product marketing, which sells software to assist customers of enterprise resource planning systems.
His company will rely upon those analytics to serve its customers, he explains, “With the Internet of Things, you've got millions of devices that can provide you additional leading and lagging indicators to give you greater visibility into your business and help drive the right actions from a business standpoint to increase your overall business performance.”
Robertson continues, “When it comes to Big Data, we're talking the three Vs, volume, velocity and variety. So being able to deal with the big data that is always associated with these millions of data sources through the IoT is a challenge.
"What’s important is the ability not only to have the level of performance that you need but also the ability to capture and integrate those data sources really effectively.”
Modeling after mobile
“The traffic patterns and scaling of IoT is reminiscent of mobile backhaul, which might provide a model for the future,” suggests Raghu Ranganathan, Ciena principal of network architecture, office of CTO, a global telecommunications equipment provider.
“In the Internet of Things, most of the traffic is from the source back up to the cloud, as opposed to the vanilla use case of a user downloading stuff from the cloud,” Ranganathan says.
Reliable connectivity is extremely important. In addition scale has to be partitioned from the perspective of there being some IoT devices that would have very low data volume and have high signalling traffic, but there could be another class of devices like CCTVs that could send lots of traffic.”
“Reliable connectivity is extremely important. Scale has to be partitioned from the perspective of there are some IoT devices that would have very low data volume and have high signalling traffic, but there could be another class of devices like CCTVs that could send a boatload of traffic.”
Ranganathan continues, “What does the network look like? Why should it be any different from the way the mobile backhaul is being done? You have things that are connected to a base station through an air interface, like WiFi or 4G cellular. You have wired backhaul.
"Organisations like 3GPP, for example, are saying ‘how do I update my LTE specifications for more upstream traffic as opposed to downstream traffic?’”
A future architecture of this type, Ranganathan adds, might look like the Cloud RAN design, which centralises and virtualises base station baseband processing.
“As a network operator, I can use my SDN paradigm to program the network connectivity to those Cloud RAN endpoints to allow the traffic to be optimally collected and processed across the network.”
Traffic prioritisation and security
A lot of applications of the IoT have been consumer-oriented - smart watches, fitness bands, home thermostats. While no service provider wants consumers to suffer outages, none of those are mission-critical.
That’s not always the case, pointed out Arie Goldberg, Omnitron Systems Technology CEO, which sells Ethernet and TDM devices.
“Some IoT connectivity services are static; once established, they will sit there forever and not change, like CE 2.0,” he says.
“There are also dynamic, on-demand services. You don’t know where they're going to pop from, whether it's a car that is travelling across the highway or whether that's some sort of pacemaker that is connecting a patient to his doctor, and he's travelling in a train.”
Goldberg continues, “IoT brings up some very interesting issues of security, reliability, especially for those kind of mission-critical type of applications, whether those are life-support type of applications, those need to be there at very, very high priority.
"It's going to be very interesting how we discriminate in favour of or against different type if IoT traffic to give priority to the ones that are more critical than others, like fire department services, instead of streaming a movie. Sometimes sufficient network resources won’t be available for everybody.”
Ciena’s Ranganathan adds, “There will be certain security embedded in edge devices, such as the ability for the device to use an IPSec tunnel. There are chipsets available that can support a secure tunnel creation. That secure tunnel could terminate at a required server endpoint, which is in a controlled environment so nobody can hack into it.”
He cautioned, however, at the limitations of that approach. “Consider smartphones. There are many possible ways of connecting to the phone, such as WiFi and cellular.
"My WiFi router at home could be very secure, controlled by me, but my LTE uplink could be a nice open gateway to come and hack my phone, record whatever, or even hijack my camera.”
The consumer, he pointed out, has no control over the security of the cellular connection.
“Security is the number one thing,” says Wedge Networks’ Zhang. “What are the potential breaches, potential vulnerabilities? In the Internet of Things, there’s data acquisition and also there are devices that cars can be driven, drones can be flying and many, many other things.
"In the middle of this is the correlation of sensor information to physical information. That means that damage can escalate very, very quickly beyond data theft into real-world harm.” That’s not theoretical, he continued, pointing out that the United States power grid has been vulnerable for years."
As a point of reference: In 2012, the US Department of Homeland Security reported 198 attacks against critical infrastructure in the United States - several of which were successful.
The role of SDN, NFV, LSO and standards
“If you look at services providers today, they've got connectivity services that already enable IoT applications,” says Anthony Peres, marketing director at Alcatel-Lucent, a networking equipment manufacturer.
“That being said, in order to broadly support the onslaught of IoT applications, there is a need to evolve to what we call cloud-era networks. SDN and NFV play a role to make these networks more agile, as well as deliver high-performance.”
Peres nailed the subject right on the head: “If you don't have visibility into the resources you have available within your network infrastructure, how can you provision a service and make sure it's actually going to work? How can you guarantee the actual performance that you need?
"That aspect of unifying service automation with network optimisation will make it faster to provision and guarantee dynamic services.”
“Not only that,” he says, “but SDN and NFV will enable the gathering of analytics to get information on what the network is actually doing, and allow for changes so that performance is ensured.”
CENX’s Timlin says: “That’s absolutely true. Of course service providers have to find value; in order to participate in the value from these machine-to-machine applications, they actually have to show more value, going beyond layer two to layer three, and really using the full capabilities of network policy control, deep packet inspection, and so-on to understand that consumer behaviour. And that's also where the analytics comes in and marrying that with the network events with lifecycle service orchestration.”
Wedge Networks’ Zhang called for standardisation to address the concerns that network operators have with the IoT ranging from intercarrier service provisioning, performance management and security. “Device manufacturers and network service providers need the IoT to be standardized so that there is no gap in between service layers.”
Zhang referenced the work of two industry organisations, the MEF, which is building specifications for end-to-end multicarrier lifecycle service orchestration (LSO), and OpenCloud Connect (OCC) which is defining standardised cloud services.
“The MEF and OCC are coming up with use cases and interoperability and compliance requirements, so that we can actually form a safe computing environment for IoT. Security breaches go for the weakest link, so that's why standardisation is very important.”
CENX’s Timlin agrees: “The IoT is creating so much volume and additional data that LSO is needed to manage virtualized network functions (VNFs).
The MEF is extending, for example, the service information model to take into account attributes for VNFs. That's really critical, because service providers are going to need to extend their capacity in a very flexible way to take into account all this volume, all the different applications. And they can only do it cost effectively with NFV.”
She adds, “The most important piece from my perspective is really taking into account virtual network functions. Also, it applies to SDN, because SDN enables the central control for programmability of VNF service chaining, so when you've got multiple services, one right after another, that also has to be taken into account for end-to-end management and orchestration.”
insightsoftware.com’s Robertson adds, “With the Internet of Things, there's a real opportunity just to be a really effective network and partner with the service providers and those cloud service providers. You're dealing with these kinds of devices. You're dealing with these kinds of schedules to optimise the network. You're dealing with mobile and stationary devices providing different types of data profiles.
"Here's how we're going to deal with it, and the kind of security that we're going to provide. And the end result is just a much more rapid adoption of the Internet of Things.”