The internet of things (IoT) is coming whether companies are ready for it or not. Most growth projections estimate that some 20 billion IoT edge devices will be connected to the internet by 2020, generating about 800 zettabytes per year. In order to better manage this data and continue to deliver low latency service to end users, companies will need to rethink their network architecture, adopting edge computing strategies and turning to edge and micro data centers.
The proliferation of IoT devices has pushed many critical processing and data collection functions to the edge of cloud computing networks. Where information once had to travel back to centralized cloud servers for processing, many IoT edge devices have enough power to handle the same tasks locally. This can greatly enhance performance, allowing these devices to respond faster to rapidly changing situations.
Hewlett Packard Enterprises CEO Meg Whitman laid out the implications of these changes during CRN’s 2017 Best of Breed Conference:
Compute and storage is going to move to the edge. The edge can be a hospital bed. The edge can be a jet engine. It can also be a factory floor. Real-time decisions are going to need to be made at the edge. You can't tolerate the latency that it takes for the data to go back to a data center and then come back to a factory floor or a jet engine. You need those decisions being made in real time.
In addition to making faster decisions, IoT edge devices are continually collecting massive amounts of data from users. Some of this data will provide actionable insights, allowing companies to make better business decisions and formulate new strategies that better meet the needs of their customers.
Michael Dell, CEO of Dell Technologies, draws a clear connection between a new generation of IoT edge devices and the growth in data traffic:
We are seeing a boom in edge computing that is driven first by embedded intelligence. When we look at the companies that make things, they are putting in sensors that are going to require all kinds of computing, [artificial intelligence], machine learning close to those edge devices. The explosion of the edge will generate way, way more data than anybody has ever imagined. All these devices, as you go from a couple of billion connected devices to one hundred billion or a trillion, you are going to generate incredible quantities of data.
Handling this massive influx of data will be one of the central challenges facing companies over the next decade. Without powerful processing capabilities on the edge of their networks, organizations face the risk of latency slowing down their services. They may also find themselves falling behind in terms of data analysis if they’re unable to develop a better solution for managing so much unstructured data quickly.
Santhosh Rao, a research analyst for Gartner, lays out the stakes:
Currently, around 10% of enterprise-generated data is created and processed outside a traditional centralized data center or cloud. By 2022, Gartner predicts this figure will reach 50%. As the volume and velocity of data increases, so too does the inefficiency of streaming all this information to a cloud or data center for processing.
While edge computing architecture provides a guide for incorporating IoT devices into high-performance networks, key physical infrastructure elements will need to be in place for these strategies to work effectively. Traditional hyperscale data centers may be useful for machine learning-powered analytics and other cloud computing services, but their distance from end users makes them less viable as outer nodes in an edge computing network. Edge data centers, which are smaller and more versatile facilities located in emerging markets, can help to empower IoT edge devices by providing processing and storage support.
Micro data centers are a specialized type of edge facility that can extend network functionality into even more difficult-to-access spaces. Often consisting of little more than a few servers in a portable cabinet or a few racks installed in a modular structure, micro data centers allow companies to augment IoT functionality anytime, anywhere, regardless of local infrastructure capacity. Micro facilities typically integrate their own cooling systems, making them incredibly easy to install and manage.
Zac Smith, CEO of Packet, sees micro data centers as a key to addressing future data processing needs:
We see micro data centers as a critical part of what’s next. While micro data centers won’t replace centralized, hyperscale deployments, most of the cloud growth in the coming decade will be driven by smaller footprints that are closer to the network edge. This will include regional sites (e.g. within a metro area) as well as far edge deployments for specific use cases.
Some of those use cases could include healthcare and manufacturing, allowing medical devices and industrial machinery to gather and transmit data with minimal latency or downtime. Combined with the potential of 5G networks, micro data centers have the potential to deliver the power of a traditional data center to even remote locations with minimal capital investment.
Smith elaborates on the ups and downs of this flexibility:
The pros are simple: they (micro data centers) can go just about anywhere and consume relatively small amounts of power and other resources. On the negative side, they are inherently inefficient to deploy and service in comparison to hyperscale facilities. This requires an entirely new approach, especially to the serviceability and automation of infrastructure, as well as deeper collaboration with the supply chain and logistics sectors.
Micro data centers are already redefining the possibilities of IoT edge devices. The ability to drop a fully functioning data center in almost any location on short notice dramatically enhances the flexibility companies have when it comes to delivering services. While they may not present a permanent solution in all cases, they can help to cover gaps in network service until 5G technology and larger edge data centers can be put in place to serve customers in developing markets. This makes micro data centers a valuable tool for companies looking to roll out edge computing strategies that take advantage of the power of IoT devices.