The impact on your network by implementing the Internet of Things (IoT) might seem fairly obvious by now. But there are some important factors you should consider before implementing the IoT in your network. The level of impact on your network is going to be dependent upon how and to what extent you deploy the IoT.
Pervasive IPv6 deployment
Obviously, since the IoT is dependent upon IPv6, you should be running a dual stack with IPv4, and be well on your way to a single-stack, IPv6 deployment. This eliminates the need to subnet, and simplifies what is going to become a much larger, more complicated network environment.
New Devices = More Data
- Internet Access — Chances are, the current pipe you have is sufficient, but with all of the traffic generated with the implementation of the IoT, it’s going to have to be bigger.
- Wireless Network — You’re introducing potentially hundreds or thousands of new clients into the wireless network. If you would prefer that its performance is not negatively impacted, you’re looking at more access points and more controllers to manage those access points. The new 802.11 standards can help here, but they are not going to make up for all of the new clients.
- Data Center — Regardless of which protocol you end up deploying, if you are bringing all of these devices into your network, it is for a justifiable business reason. Each new device is going to be transmitting data into the network, which then needs to be stored and analyzed. Chances are, no matter how much future growth was built into your last data center upgrade, it won’t be nearly enough. This is why if you speak to most people involved in technology sales, they see the data center as one of the greatest potential areas for growth. It may be time to consider that hybrid data center/cloud solution you’ve been discussing for the last few years.
- Performance Monitoring — The way you monitor performance is going to change to accommodate this new data. If not properly planned for, the massive amounts of new data stored on your network is bound to create performance issues. If your performance monitoring platform does not scale well, you could find yourself monitoring certain things less frequently, monitoring only those things you deem most important, or using average metrics to produce capacity forecasts, which presents problems down the road. Monitoring platforms will now be measured by speed at scale. In other words, as the size of your network increases geometrically, your monitoring platform is able to perform at the same speed as previously needed.
Irregular Traffic Generation
These new devices we’ve introduced are going to generate traffic in short bursts on irregular schedules. That is, only when they have to. Polling your network at five-minute, or even one-minute, intervals is no longer going to work. It may be necessary to poll every second to allow for the billions of devices that will be communicating over the IoT.
Short-burst traffic like this, when monitored at five-minute intervals, may be flattened out. This could create problems with latency-sensitive applications like VOIP and video. The difference between polling your network at five-minute intervals versus sub-one-minute intervals is like the difference between a standard DVD and a Blu-Ray disc. All of those imperfections you see on a high definition disc are not visible on a standard disc.
If you want a true view into your network’s performance, make sure your monitoring platform supports high-frequency polling. This will make troubleshooting performance issues in the IoT world much more efficient and effective.
Increased Cloud Dependency
We’ve discussed the potential impacts of the IoT on your network. What now? Do you make that difficult trip to your CFO’s office and present him with next year’s IT budget, which would likely result in a trip to the cardiac unit of your local hospital? Or maybe you need to make that presentation to the local planning board to convince them that if they really want to keep up with the school system in the town next door, that everyone’s property taxes are going up.
If those are things you’d rather not do, you’re likely to consider other options, the most obvious one being transferring some of your IT operations to the cloud. Of course, surrendering even a small part of your operations to the cloud limits the amount of control you have over that part of your operations.
The National Institute of Standards and Technology’s (NIST) working definition of cloud computing cites five essential characteristics of cloud computing:
- Broad Network Access — Capabilities are available over the network and accessed across a varying set of platforms or end-user devices, such as mobile phones, tablets, laptops, etc.
- Resource Pooling — The provider’s computing resources are pooled to serve multiple consumers using a multi-tenant model, whether we’re talking about storage, processing, memory, or network bandwidth.
- On-Demand Self-Service — A consumer can unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service provider.
- Rapid Elasticity — You have capabilities that can be rapidly provisioned and released to scale in a manner that’s commensurate with demand.
- Measured Service — Clouds leverage a metering capability at some level, so resource usage can be monitored, controlled, and reported, which provides transparency for both the provider and consumer of the cloud service.
These five areas provide an initiation point for the discussion. You may already be using some level of cloud services. How much more of your operations you want or need to turn over to the cloud is both a business and technical decision.
ABOUT THE AUTHOR
Rich Hummel has over 30 years’ experience in different areas of technology. He has degrees in electrical engineering and management, as well as a variety of technical certifications.
This is an excerpt from the Global Knowledge white paper, The Internet of Things: A Primer for the Curious.