Gartner estimates that by the year 2020, there will be approximately 20.8 billion things connected to the Internet in addition to the traditional connected items such as computers, smart phones, and tablets. (Gartner, 2015) That’s about three times the estimated number of people who will populate the planet in 2020. (UN Data, 2017). For enterprises that sell, deploy, manage, and use IoT devices, this will increase the demands placed on their current network infrastructure. In order to meet these demands, IT leaders and practitioners need to rethink a fundamental components of their application and network architecture including DNS.
Let’s use a connected washer/dryer as an example. This particular washer/dryer is located in New York City making it much more efficient, from a latency perspective, for it to get its software updates and connect through to end user devices from data centers located on the east coast of the US, in this case NYC1 located in lower Manhattan. Similarly, if the washer/dryer was located in Los Angeles, it would be better to route it to to data centers West Coast. Geographic routing ensures that IoT devices are routed to the closest physical data center which may be more performant than more distant locations. Of course, if the geographically closest data center were to go down for some reason you would want to reroute traffic away from the outage and to the next best data center.
For the purposes of our example, let say a major hurricane has hit the New York City area flooding lower Manhattan - where NYC1 is located - and taking out the power. To make matters worse the lower levels of the building where the backup generators are located are under water (this actually happened to a number of data centers located in lower Manhattan during hurricane Sandy in 2012).
Using plain vanilla DNS, IoT devices on the eastern half of the US would continue to be routed to the New York City data center, which is hard down. But by using an intelligent DNS solution, that IoT traffic is automatically re-routed to the next most performant data center, minimizing the impact of the outage. This scenario is called automatic failover and is considered a best practice for application developers who need to keep their applications online event when the unexpected - even catastrophic - happens.
Being prepared for the worst is critical, but proactively working to ensure IoT devices stay connected is equally important. Enter load shedding. Let’s assume the storm has cleared, the waters have subsided, and the power is back on in New York City. Now the issue is that a spike in traffic is sending servers in NYC1 near their maximum capacity. NYC1 may still be the optimal destination for east coast traffic, but, in this case, you would want to send traffic to the West Coast data center to ensure you don’t overload NYC1’s servers. Load shedding allows application developers to leverage multiple data centers, ensuring the washer/dryer is always connected and protecting against data center outages due to traffic spikes.
IoT requires enterprises to invest in new network and infrastructure technology to support the millions of connected objects that are coming online each year. This trend is not going away and will likely accelerate as new use cases and new market opportunities for IoT are discovered. Just like laptops and servers, IoT devices utilize DNS to connect to endpoints across the Internet. That is why IT leaders and practitioners need to include intelligent DNS & traffic management in their IoT roadmap.