The Internet of Things (IoT) may be the biggest thing to hit the data center market since the cloud. Consider that there are billions of devices on the planet that use sensors; also consider the growing number of sensors that can send data over the Internet. Gartner, Inc. estimates that there will be 5.5 million IoT devices in place this year, and that by 2020, there will be 20.8 billion, with an associated revenue of $300 billion, mostly from services.
While industry experts believe that consumers will generate the most IoT data traffic, the enterprise will do most of the spending on building an IoT infrastructure. A recent Gartner forecast predicts that the consumer sector will see about 4 billion things connected to the Internet this year, but that vertical-specific smart devices will hit 1.25 billion devices. The number of cross-industry devices, such as light bulbs, also will boom, hitting 1 billion this year.
What this means for data center architects is that they now have to deal with a flood of machine-generated IT traffic, including determining how to prioritize, route, process and store IoT data. A new burden is being placed on the enterprise infrastructure, and what may be a trickle today will grow into a flood of IoT information very soon.
What’s the best way to prepare a data center for the coming of IoT? Here are seven considerations:
1. Data storage management. IoT will generate a lot of data, so data storage and storage management are going to be primary concerns. As the number of sensors grow, so will the amount of IoT data traffic. Sensors inside and outside the enterprise are going to provide petabytes of data that may someday prove useful. Therefore, the initial focus will be on storage capacity (especially cloud storage) and determining how a business gets the most value from incoming IoT data in a cost-effective way.
2. Cloud data storage. Because the anticipated amount of IoT data is beyond the storage capacity of the enterprise, data center design will to have to include more cloud data storage. When considering cloud storage for IoT data, there are some added factors to take into account :
• Most devices are often connected directly to the storage facility. With a direct link to the cloud data repository, data transfer is faster, so less storage space is required on IoT devices, which saves on costs.
• Storage management now becomes the cloud provider’s problem. Service-level agreements and cloud provider guarantees become more important. At the same time, storage management becomes less of an IT management headache.
• Cloud computing is ideal for IoT data processing. The value of IoT will be revealed in big data analytics, which will require cloud resources.
3. Security. Capturing IoT data will raise new security concerns. IoT data is often sensitive and proprietary. Maintaining a direct connection between the IoT device and the data repository is efficient but also risky. There are more opportunities for a security breach. The good news is that most cloud providers have strong security systems, but there’s always human error. Strong security protocols will be required.
4. Big data. Big data analytics will create additional security challenges, mostly due to security complexity. For example, real-time processes will be hampered by data availability requirements, creating a risk to real-time operations.
5. Bandwidth. Most data center infrastructures are scaled to handle moderate wide area network (WAN) traffic, such as data, unified communications and applications. IoT is going to flood the pipe. Now the data center network will have to cope with new traffic patterns, accommodating massive amounts of small messages from IoT sensors for processing. This means data centers will need more inbound bandwidth.
6. Server technology. Even with virtualization and cloud computing, data centers are going to need more powerful servers. For example, a new server has to be added for every 600 smartphones in use today, so more servers and server capacity will be needed in order to handle IoT traffic. What’s worse, the data traffic will translate into diverse workloads that are difficult to predict.
7. Edge computing. International Data Corporation expects workload demand from IoT to increase 750 percent by 2019. Servers and storage capacity won’t be able to keep pace with demand, so expect to see more hyperscale data centers handling IoT processing at the edge of the enterprise. More edge-based data processing will significantly reduce IoT traffic, and data also will have to be encrypted at the edge for protection.
These are just a handful of changes to consider when designing a data center to handle IoT traffic. Be assured that the flood of IoT data is coming, and it’s up to you to help your customers keep the flood at bay. More servers, more data storage and more cloud resources will be needed, but what they will need most is more knowledge and a better understanding of how the IoT is going to change their data center design.