Data is vitally important for businesses today, supporting decisions ranging from optimizing production in real time to predicting (and avoiding) equipment failure. Many companies gather and analyze this data through IoT deployments that consist of sensors and gateway devices—the sensors take in data and send it to the onsite gateway device, which then processes and performs initial analysis, and, in turn, sends the relevant data on to the cloud.
Traditionally, IoT deployments rely on embedded computing architectures, which means that the software is tightly integrated with the hardware it runs on. The problem with this situation is that making any changes to the software can be difficult because of how intertwined it is with the underlying hardware. Updating, adding, or deleting applications is therefore not easy, and often requires you to physically access the gateway in question. For IoT deployments in remote areas—wind turbines, oil rigs, etc.—each in-person visit can be enormously expensive.
Faced with the need to update their gateway software, most businesses instead typically decide to “rip and replace” the entire system, removing their old gateway device and installing a new piece of hardware with the latest applications included. However, as soon as changes need to be made to this new software, the cycle repeats itself. It’s an inefficient, expensive, and wasteful approach to IoT, but for a long time, it’s been the only viable option.
Now, modern technologies like virtualization are being applied to edge devices and are disrupting the status quo of embedded computing. By virtualizing the underlying hardware of edge gateways, it’s easy to add, remove, or update the applications on the device because they are no longer tied directly to the hardware.
An additional benefit of virtualization is that it allows applications on the same device to run in complete isolation from each other, which makes it easier to host multiple applications on one gateway. If a business needs to add or change the software on their gateway, they no longer have to buy a whole new device to accomplish that task, or even visit the site at all, saving time and money.
What’s interesting about the old “rip and replace” methodology is that it was necessitated by the embedded application architecture, rather than actual physical device limitations. In other words, edge gateways have long had the capacity to run multiple applications simultaneously and to reconfigure them if needed—they simply haven’t had the software architecture in place to do so. To use an analogy from chemistry: the hardware isn’t the rate-limiting factor.
You may wonder: in what circumstances would a business want to adjust the applications running on their edge gateways? There are actually many use cases for making changes. For instance, Zededa works with a large wind farm operator in North America. Currently, this
company uses sensor data to detect when the machinery of the turbines begins to operate abnormally. The sensors send this data to an edge gateway, which analyzes it for anomalies—readings that deviate from what is expected—and then issues an alert if the data indicates that maintenance may be necessary, before the issue escalates to the point that it causes the turbine to be taken offline for a major repair.
Recently, this company has begun exploring the idea of adding sensors to detect weather and force majeure events that may damage turbines, too. This new type of sensor data will require a new application to process, analyze, and issue alerts about the data when needed. Under the old embedded system, the company would have had to add a new, second gateway with the weather detection application installed to each of their turbines—a second gateway for the second application.
However, by virtualizing their gateways’ hardware, this company is able to use application orchestration to deploy an event detection application on their existing devices, without interrupting the preventative maintenance application’s workflow. It will save them significant resources and avoid the logistical difficulty of installing new gateways within each turbine for each new application.
Virtualization also provides additional benefits beyond making it easier to reconfigure the software on gateway devices. One example is making over-the-air (OTA) firmware and software updates simpler and more reliable. Traditional gateways with embedded architectures can sometimes be updated OTA, but the challenge of this approach is that there is no rollback option in the event that the update is faulty.
If something goes wrong, the whole device could be bricked, meaning that you would have to send a technician on site in order to restore functionality, costing time and money. When devices are in difficult-to-access locations, this can be particularly challenging!
How does virtualization solve this problem? The short answer is: by separating the software from the hardware. For example, if you need to update a gateway’s operating system, the update can be downloaded and installed OTA in a separate virtual machine (VM) from the rest of the device’s applications. The new update can then be tested and confirmed as good in this isolated area before being applied to the whole device. In the event that the update is faulty, it’s simple to roll back to the last “known good” state.
We’re still at the beginning of the possibilities that IoT can create for businesses, and technological advances like edge virtualization are helping to unlock new opportunities. Traditionally, computing equipment was placed in the field and depreciated in value over time, until it was literally thrown out. There is even a term for this: “set-it-and-forget-it!”
Now, new applications can be installed on existing hardware easily and remotely, with each new application providing new, valuable business insights and actually extending the ROI of the hardware well beyond its previous lifespan. It’s also much simpler to patch and update applications when needed with virtualization, meaning that IoT deployments can be consistently upgraded. It’s an exciting time for enterprise and industrial IoT, with the ROI for hardware increasing as new applications to extract value from data are developed.
