In recent years there has been a shift away from the typical linear expansion of networks in line with the number of users – one user once meant one PC or laptop, but with users becoming increasingly mobile, and increasingly connected, that same user today may also carry a tablet, smartphone or wearable technology – significantly increasing the active user devices on our networks.
Even today this presents real challenges for maintaining a consistent user network experience across devices and locations whilst securely mobilising our enterprise network services beyond the desk, and outside the office. But this increase in scale is just the tip of the IoT iceberg. IoT takes things dramatically further as it decouples network devices from the hands of a user, introducing machines and sensors into our networks to collect data and analytics to drive smart outcomes.
Networks cannot be focused only around the human and our end-user devices any longer.
Even the most conservative analyst predictions for the growth in Internet-connected IoT devices are staggering, with the number of connected devices set to explode from some 10 billion globally in 2016, to estimates typically around 25 billion by 2020 – that’s a (conservative) rise of over 300 million new IoT connected devices every single month… and some analysts put the figure significantly higher still!
With each of these devices also transmitting data over our networks, Cisco estimate that the monthly volumes of new data transported and stored will quadruple between 2015 and 2020!
This explosion in the number of IoT devices connecting to the network, and the new traffic, as a result, will create unprecedented pressure on our network and storage, requiring new infrastructure methodologies to support the data analytics, insight and smart decision making that IoT can deliver.
Even with our many smart devices in our pocket, Cisco analysts anticipate that the data transmitted by IoT and M2M devices will be a staggering 269 times greater than the traffic generated by human operated end-user devices by 2020, and this requires a new approach to installing networks that can scale quickly and easily to accommodate new devices and traffic flows.
Our IoT capable networks must be truly multi-service capable of supporting new data flows concurrently, with effective isolation and prioritisation to keep concurrent services running smoothly and efficiently from end-to-end across the network. It should offer the flexibility to cost-effectively scale up and scale out, to add and efficiently utilise new network paths across an elastic infrastructure that can easily accommodate new wired or wireless capacity whilst avoiding costly and disruptive fork-lift upgrades.
The world’s commonly installed, but now legacy network technologies cannot easily support this transition towards an IoT connected world and should be assessed for their suitability before an IoT strategy is adopted.
Legacy networks are built on traditional, static layer 2 topologies, with VLAN’s and Spanning Tree protocols used to provide rudimentary segmentation of traffic and deliver crude network rings to support a base level of redundancy, but neither technology scales to accommodate large volumes of network devices. There are hard limits to the number of VLAN’s that can be run on switches, and practical limits to the number of devices that can reside within a VLAN, especially if devices are to connect over WiFi. On top, we have the challenge of Spanning Tree – an unbearable ‘resiliency’ protocol that often creates outages on networks as it recalculates the available paths through a
network following a topology change, and the timed require for recalculation increases as the size of the network grows, yet in traditional networks there are no ubiquitous alternatives.
To overcome the problems with layer 2 networks, layer 3 routing is often used to transport traffic over the backbone of the enterprise network where we can avoid Spanning Tree, but this is often architected in an open-access framework, with traffic able to hop between ‘isolated’ VLAN’s as soon as that traffic joins the routed network. Organisations that rightly use policy on their Firewalls to restrict this open-access between VLAN’s face the new problem of maintaining an increasing number of access lists between their new IoT networks, and risk a substantial hub-and-spoke traffic bottle-necks as the Firewall must inspect data from many new sources.
I’ll talk more about the headline grabbing security challenges of IoT in a separate article, but even at the connection and backhaul level IoT creates problems for organisations seeking to gain benefits from the technology.
New network technologies have evolved in recent times to overcome these scale challenges of traditional networks, and should be considered by organisations that are investigating IoT and M2M technologies. Shortest Path Bridging (SPB), IEEE standard 802.1aq, supports the easy deployment of efficient, flexible, meshed networks, removing the need for Spanning Tree on the network backbone, and allowing all links to run active-active. IoT networks can be isolated end-to-end without the limitations of VLAN’s, and these networks need only to exist at the edge, rather than spanned across all devices in the network. With load-balancing capabilities built-in, the SPB protocol efficiently steers traffic onto the shortest path to its destination, helping to avoid the bottle-necks of a hub-and-spoke network and creating an infrastructure that supports easy expansion to add new switches and new IoT flows here, there and everywhere across the Wired and Wireless network.