Why IoT needs Software Defined Networking (SDN)

Software defined networking (SDN), with its ability to intelligently route traffic and take advantage of underutilized network resources will help stop the data flood of IoT. Cisco has a pretty aggressive IoT strategy, and they place their application centric infrastructure version of SDN at the centre of this. And it makes sense. Software is still the main ingredient that can be used to combat network bandwidth challenges.

Lori MacVittie⁸ agrees with SDN being a critical enabler, but only if SDN considers all of the network layers from 2 to 7, and not just stateless 2-4. “Moving packets around optimally isn’t easy in a fixed and largely manually driven network. That’s why SDN is increasingly important when data volumes increase and do so in predictable waves. SDN can provide the means to automatically shift the load either in response or, optimally, in anticipation of those peak waves.”

The network challenges in IoT do not stop at bandwidth and volumes of data. Applications will be required to deal with the peak loads of data, so services will be required in layers 4-7 that provide for scale, security and performance of those apps.

SDN has features that will also be particularly useful. Dynamic load management should allow users to monitor and orchestrate bandwidth automatically on the fly, which will be music to the ears of global IoT providers. Service chaining will enable application specific processing procedures in a sequence fashion to a client’s job. This should ease management overhead in IoT services, as the subscriptions increase globally. One of the coolest features of SDN is bandwidth calendaring which will allow the user to schedule the traffic an application will need at a given time, and when you think of a sensor only wanting to communicate at periodic times, it is apparent that this will be a great asset.

But this cannot happen soon. Data center managers will have to modernize their infrastructures. Once they do, a potential big win would be the ability to create numerous virtual and private networks on top of a single physical network. This would be a big advantage as multiple customers could then share a single network, without risk for their applications and data. However, for this to work, one would need the entire network to be SDN enabled.

When one considers the concept of Network Functional Virtualization (NFV), this path can be traversed quicker. With NFV ready networks, carriers can create services in software, rather than dedicated hardware, essentially allowing virtualized servers to allow these new services. This enables business transformation by moving away from having multiple isolated networks, and one would work with an open ecosystem, a set of virtualized network functions, and most importantly an orchestration layer. This will allow businesses to accelerate with agility in the face of device quantity explosion.

Reference:

8: Dev Central: SDN and IoT article

https://devcentral.f5.com/articles/sdn-is-important-to-iot-if-it-covers-the-entire-network

Considerations of Change: An Intro to Networking in IoT

One of the major consequences of Moore’s Law for silicon is that pretty much any device now can have a reasonable level of computing power and internet connectivity. Because of this, the number of internet enabled devices is increasing, thus causing a huge influx of IoT traffic; it is predicted that WAN bandwidth will need to be increased.

When one considers the types of data that will be generated, it becomes clear that they both present challenges. George Crump, an analyst with Storage Switzerland points this out⁷. “First, there is large-file data, such as images and videos captured from smartphones and other devices. This data type is typically accessed sequentially,” explains Crump. “The second data type is very small, for example, log-file data captured from sensors. These sensors, while small in size, can create billions of files that must be accessed randomly.”

From this, it is clear that data centers will need to handle both types of data, and the storage and processing requirements that come with them.

For decades, the network was considered to be the plumbing of a company’s IT solutions, and was considered a somewhat dumber element of the design. With the advent of IoT, it is clear that the networking element of the IoT ecosystem is slightly lagging behind, which is a concern as IoT is very much a network centric technology, and in essence makes the web by which the sensors communicate to the host and to each other. There are a number of ways for these devices to be networked. Some devices can be directly connected to the internet utilizing standard Ethernet or Wifi, which are TCP/IP based. There are other wireless technologies, some of which are dependent on TCP/IP, but all require some sort of intelligent gateway to convert their network into standard Ethernet or Wifi. These include, but are not limited to, Zig Bee, Z-Wave, Bluetooth, and Cellular.

Evolution towards IPv6 

Due to the advancement of object gateways, the first two stages of the IoT roadmap will sit on current infrastructure and protocols. Once the volume of devices and data increases and true IoT is in motion, the IPv6 protocol will be required, which offers unlimited IP addresses.

The main challenge that IPv6 looks to overcome is the large packet size when we consider standard IP protocols. For IPv6, the packet size is reduced by making a number of changes to the release of the 6LoWPAN standard, namely RFC 4944. Changes included the compression of IP headers and the introduction of a fragmentation mechanism that enabled reassembly of IP packets that did not fit the IEEE 802 packet. Lastly, routing protocols for lossy, low power networks were required. New protocols were developed by the Internet Engineering Task Force (IETF) that provided basic routing in low power lossy networks.

In my next blog post, i will continue to write about network enablement requirements, talking about why IoT needs “Software Defined Networking” (SDN)

Reference:

7: Orange Business: Can your business handle IoT

http://www.orange-business.com/en/blogs/connecting-technology/data-centers-virtualisation/can-your-data-center-handle-the-internet-of-things

IoT and Classical Business Models

Many companies, especially in the Information Technology (IT) section are aware of the IoT explosion, one of the biggest challenges facing any company is how they prepare for the change that will result from the increased business impact that IoT will present.

With figures in the trillions in terms of the market for IoT, how do companies ensure they can get a slice of the pie? If they currently do not sit within the relevant market segment, analysis will be required to determine if it can be an opportunity or a threat to their business as a whole.

IDC in 2014³ predicted that IoT will actually overtake the Information Communication Technology (ICT) over time. It predicts IoT will grow 12% year on year, whilst classical ICT will grow just 4%. Figure 3 below illustrates this.

Considering that most business is consistently monitoring the bottom line, it is not only the opportunities that it will present, but how it will impact how we work. With limitless numbers of sensors monitoring processes, improving business energy efficiency, enabling new ways of working in teams, business will need to be more open to change, and more dauntingly, open to the elements of a “big brother” type scenario.

There are trends that are ensuring an evolution of business practice as we know it. Normally, new technology platforms impact on a single strand, with the exception of the impact of the internet. But IoT has the potential to become an entire business ecosystem, where creating and capturing business value will be paramount. However, this is not a straightforward suggestion. Barriers to this include the current early position of IoT in its lifecycle, and the sheer volume and types of devices to be considered. From an ecosystem perspective, by nature it would indicate a seamless quantity of micro-systems working together in a self-sustaining fashion. Trying to estimate what this will mean for IoT is still not clear.

Consider the classical technology adoption lifecycle. There are five types of innovation adopters, the first being the innovators themselves. The list is completed, in sequence by early adopters, early majority, late majority and laggards. With the current immaturity in IoT, and the lack of clarity in the various emerging technologies, the challenge for business is to try to advance the early adopters to early majority, so the business needs to be able to scale. The early adopters are less fussy when it comes to product design, but once the number of adopter’s increases later in the life cycle, the early majority will want polished product offerings, with appropriate services.

With the IoT still in its relevant infancy, it is appropriate to compare it to the early stages of the Internet. When we look at the recent business ecosystems that have been spun out of the Internet for EMC, such as Pivotal Cloud Foundry, one would postulate about future ecosystems opportunities for EMC from the IoT spectrum.

Another important consideration for companies is to consider the skill-sets and people that are required to drive their Big Data strategy as a result of their growing IoT ecosystem. A key tenant for this will be the data itself, and in the February IT@Cork Tech Talk by my EMC colleague Steve Todd, and even more recently in his blog on data value (value was something I had never associated to data until this talk), Steve spoke to the importance for major companies to begin to consider a more structured approach to their employees that are involved in data set discovery, identification and migration (Data Architect) and also a Chief Data Officer to represent the company from a data perspective. Interestingly, my role in EMC changed last year, to the role of a Data Architect. So I could first hand relate to this. When faced with a business challenge in big data, 5 steps that can be critical to success are as follows.

1: Demystify and then map the current devices, tools, processes and trajectory of data across the business unit or company (AS-IS Diagram)

2: Scour the company and external sources for any technologies that can enable a more scalable and clearer approach

3: Look to centralize data storage, to allow the company to focus on being agile and scalable, and also remove duplicate data (concept of a Business Data Lake)

4: Develop an ingestion framework to ensure the data lake has a sufficient landing platform for data.

5: Build the Analytic’s platform that is pointed at the centralized “Business Data Lake” to meet existing and future needs of the business.

When we apply this to IoT, we start to that every company, no matter how small, will begin to generate huge data-sets, and there will be a new skillet needed at companies that never had previously to ensure they can gain as much insight from the data sets. Sure, there are companies that can provide these solutions, but realistically, the future state will surely be to have these as core skills, just as “internet skills” once appeared on resumes?!

It is proposed here that key stakeholders across multinationals can overcome these challenges and design practical IoT business models if they consider an ecosystem style approach, instead of looking at modular needs of individual business units. This will allow the business to get a high level perspective of where IoT can bring value to their business offerings.

Reference:

3: Digital Universe Article

http://www.emc.com/leadership/digital-universe/2014iview/internet-of-things.htm

An IoT Data Flood. Are we ready? (Intro)

A flood of 50 billion pieces. That’s the predicted number of internet enabled devices that will span our globe in 2020 to create the expanding Internet of Things (IoT). And it is a conservative estimate, when you consider other types of technologies that could be enablers, namely Near Field Communication (NFC) and Radio Frequency Identification (RFID). The speed of internet connectivity in the future is likely to hyperscale, and makes Moore’s Law, which we successfully navigated, seem tortoise like.

So what will all this mean? Crop fields will be smart. Crime will diminish. Stagnant business models will become fluid. Heck, we may even predict the weather. But one thing will remain: The need to collect, store and analyze this data. The coming years will see a dramatic and disruptive innovation of the classical data center model as we know it. It will not be practical for all these remotely distributed devices to transfer their data to centralized data centers. In recent years, data center consolidation has accelerated, yet it does not fit well with IoT. It is proposed in this article that a single person’s life and home of tomorrow will generate more data than the industrial plant of today.

Whilst estimating the impact of the Internet of Things (IoT) over the coming decade would be difficult to do accurately, one thing is apparent. It is going to be a game changer. With the number of devices , the ability to connect, communicate and remotely manage these automated devices is becoming an enabler, from the parking lot to the factory floor to the homes we live in.

A critical enabler for IoT longer term is the concept of smart cities, where both human centric wearables and machine sensors will work together to make the cities of tomorrow more efficient, secure and safe. By 2050, it is predicted that two thirds of the world population will live in cities. This migration naturally represents great challenges especially in healthcare, security and energy use.

Sogeti², a global collection of over 120 technologists, makes an excellent association between smart cities SMACT (Social, Mobile, Analytics, Cloud and Things) and the concept of a platform. The City as a Platform is twofold: it is the infrastructural capacity plus the human dimension, the empowerment of behavior via data and applications. It shows that the digital architecture of a city is beginning to look like a platform with various abstraction layers that support one another. There are 11 scenarios in which a city can become smarter: waste, healthcare, grids, retailing, supply chains, tourism, e-government, smart meters, food, traffic and logistics management.

From Figure 2 above, the top shows the activities of everyday life, with citizens, students, consumers and commuters. Below this is an abstraction layer containing technology such as an Application Programming Interface (API). Streets become smart if we can link camera systems with facial recognition technology. As you traverse the layers, you will notice common elements of any platform, with communication and/or collaboration between these layers. These are already in action, apps like Air B&B and Halo/Uber show that smartness in applications can make cities more efficient in regard to transport and space respectively.

Many people own internet connected devices, such as their smart phone, laptop or smart TV, but this is the beginning of an age where technical advances and cost reductions mean elements such as baby monitors, fridges, temperature sensors, in-home heating and lighting will all be connected. The list of devices is growing all the time. But if we stop and think about what these devices mean for the classical data center model, it soon becomes apparent that this deluge or flood of data will impact data storage, processing and analytic platforms that we use today.

The strain is evident already, and that is with the devices that we control (laptops and phones generating data by surfing the net for example). It’s still just two devices per day per person. Imagine if that number increases over 50? Imagine the data load and bandwidth implications when those devices are sending data regularly? Then consider an entire city of people with the same level of internet connected devices, leading to billions of devices generating vast quantities of data which must be processed and stored. Understanding this impact is important if you are to ensure that your infrastructure is correctly designed to support an IoT strategy that your organization will need to remain competitive in the coming decades.

In my next blog post, I will explore the impact of IoT on Classical Business Models. Stay tuned!

References:

1: Connectivist Chart on IoT Growth

http://www.theconnectivist.com/2014/05/infographic-the-growth-of-the-internet-of-things/

2: Sogeti Labs: City as a Platform Article

http://labs.sogeti.com/internet-things-cities-platform/