macrovector - Fotolia
Edge computing and IoT combine to reshape the network
Enterprises are examining edge computing and IoT in a new light to determine the best way to process and transmit the mountain of data generated by IoT devices.
Without question, IoT is reshaping the network. The sheer number of gadgets and devices now being installed -- from simple smart speakers to sophisticated medical imaging scanners -- generates reams of data, all of which has to be moved and processed.
As a result, enterprises are paying attention to edge computing, which places compute, storage and processing capabilities closer to the devices that pump out the data. The big reason why edge computing has garnered attention is it's more efficient and cost-effective to deal with the data at the edge of the network than to transport it all to servers at a centralized data center.
Consider an AI application that can recognize anomalies in a production line by analyzing video captured in real time. That video creates a lot of data -- especially if the video monitors many observation points in the production line. To move all those bits and bytes across a WAN -- whether via public internet or private -- to a central data center would require a lot of bandwidth. It would take time, and time equals delay. Any delay in spotting a production issue can have a significant effect on quality and revenues.
Or take an app focused on site security. In order for that software to be useful, the network would have to be able to crunch data quickly and send out alerts immediately to enable the security team to respond.
One of the biggest benefits associated with edge computing and IoT is reduced WAN transport costs. Keeping the data local also prevents quality of service issues that could result if a deluge of IoT data collided with more typical business transaction, VoIP and other traffic. By keeping data at the edge, processing can take place immediately. Any issues or key data analysis required is available sooner and, as a consequence, alerts or summary data can be quickly transmitted to experts at a central location.
Making the edge computing transition
That was the good news. The bad news is few of our networks have been architected with this new model in mind. Prior to the advent of IoT applications, edge networks generally handled low-volume, low-speed users. A typical implementation was either a desktop connected to Gigabit Ethernet (GbE) or a laptop or mobile user connected via Wi-Fi.
High-volume IoT devices will demand more bandwidth, as will server connections in your data center. While you don't need to toss out your existing edge networking gear, you will likely need supplemental, upgraded equipment to support the edge computing and IoT elements at your edge location.
Here are some edge computing and IoT trends to track, beginning at the edge of the edge and terminating at the edge computing center -- i.e., the data center.
Wi-Fi 6 and multigigabit. Wi-Fi has always traded speed for convenience. Until recently, wireless LAN access points (WLAN APs) topped out at speeds well below 1 Gbps. Wi-Fi 5 -- 802.11ac -- broke the GbE barrier. Wi-Fi 6, which will be formally ratified later this year, will push aggregate throughput beyond 10 Gbps.
As a result, new and next-gen APs are well-positioned to support a variety of IoT implementations -- from video cameras to large numbers of sensors. Usually, this data is all upstream and destined for the edge computing data center. But to get there, it will need to move from the AP to the wired network. If that connection is a single GbE link -- as has typically been the case -- the AP can become your edge computing bottleneck.
To bridge the bandwidth gap, standard-setters and vendors in 2016 put together a new specification, which enabled copper-based 2.5 GbE and 5 GbE connectivity. This spec gave APs the bigger pipeline they need to connect to the wired campus network, as long as both the switch and AP support multigigabit.
Vendors today offer switches equipped with a number of multigigabit ports in order to support multiple APs, rather than just adding one or two multigigabit ports to existing GbE switches. If your edge data is going to be arriving via WLAN, look at Wi-Fi 6 and multigigabit to avoid inbound bottlenecks.
Edge data center. Hyper-converged infrastructure (HCI) and 25 GbE links are superseding the days when 10 GbE links were sufficient enough for data center LAN connections used to bridge compute servers to storage resources. HCI simplifies the data center by combining compute and storage into the same hardware infrastructure. Multinode systems enable both scalability and failover, or robustness.
Edge computing will be a requirement for many companies as large-scale IoT deployments take place. Bring Wi-Fi 6, multigigabit Ethernet, HCI and 25 GbE into your edge computing and IoT plans now so you will be sure to be ready.