skvoor - Fotolia
More IoT devices mean more IoT data -- which means more organizations that need to process and analyze that data more quickly.
Until recently, companies turned to the cloud to process and analyze massive amounts of data. But when it comes to IoT, the cloud is not enough. Enter edge computing. By processing and analyzing data as close to the originating source as possible, edge computing in IoT is set to be the norm.
So what IoT issues does the edge solve, exactly? Here are six of them.
1. Latency issues
Latency, in the context of this article, refers to the time it takes to process and analyze captured data. A device that's connected to the internet has to respond in less than 100 milliseconds -- or sometimes even less than 10 ms. Consequently, compute assets must be as local as possible to offset the inherent latency in transmitting data over distance, said Christian Renaud, a research director at 451 Research.
"That's one of the key reasons cited for edge computing. It's right there, it's a very quick feedback loop and that immediacy is what many applications demand," Renaud said.
The greater the distance between where data is created and where it's collected, processed and analyzed, the greater the negative effects on this data, such as the ability to process it at the necessary speeds.
With edge computing in IoT, the computing is done close to the source -- for example, the sensors -- so there are no bottlenecks in terms of transporting data from one place to another, said Aapo Markkanen, an analyst at Gartner.
"For example, if the sensors in a car figure out that there's going to be a crash, then the system has to be deterministic enough to deploy the airbags in a certain time frame," Markkanen said. "If there was any [lag] in the transmission of data over longer distances, it would not be safe at all. That kind of stuff you have to do right at the edge."
2. Bandwidth issues
Most IoT devices that run software and generate data need to be linked to the cloud to store and further process that data. As such, massive amounts of power and bandwidth are needed to transmit IoT data to the cloud.
Edge computing in IoT allows organizations to reduce their internet bandwidth usage as large amounts of data can be processed near the source.
"The more you do at the edge, the less you have to transmit over distances," Markkanen said.
For instance, edge-computing cameras could help law enforcement agencies reduce bandwidth by analyzing video feeds from police dashcams, which generate immense volumes of video and audio recordings, in real time but only sending the relevant data home when necessary.
3. Bandwidth cost issues
"IoT applications generate large volumes of relatively low-value time-series data, in essence 'spamming' the cloud with frequent small updates," Renaud said. "And that means there's the cost of bandwidth, the opportunity cost of the equipment to get to that bandwidth, the storage and the analysis costs, and the compute costs in the cloud of this low-value time-series data."
With edge computing, this data can be captured and -- if necessary -- analyzed as well as summarized before it's sent to the cloud or another upstream aggregation point, he said, which is much less expensive than sending unfiltered data over WAN links, which are often very costly.
4. IoT security issues
Although cloud providers have developed excellent security for their IoT offerings, operational technology professionals are still concerned that their sensitive data will not be safe once it leaves the walls of their enterprises, Renaud said.
To solve this, Markkanen added, more intelligence can be added at the edge to secure systems, making them stronger against hacks and intrusions.
"If I have more powerful edge devices, I can update my security on those devices," echoed Holger Mueller, an analyst at Constellation Research. Consequently, any outage would be limited to the edge computing devices and the local applications on those devices.
5. Data sovereignty compliance issues
Having every device connected to the cloud and sending raw data over the internet can lead to privacy, security and legal implications, particularly when it comes to sensitive data subject to the regulations of different countries.
Many nation-states and other government bodies are reluctant to share sensitive IoT data outside of their sovereign boundaries, Renaud said. By processing data close to the source, organizations can keep the data within their borders and ensure compliance with data sovereignty laws.
The German Bundesdatenschutzgesetz data protection act and the European Union's GDPR are examples of national regulations that stipulate what and where data can be exposed, processed and stored.
The overarching aim of these regulations is to give citizens more control over their personal information. Implementing edge computing in IoT means that data can be stored in local on-premises data centers, guaranteeing control over and restricting access to this data.
6. Legacy-system bridge issues
The legacy systems that companies often connect to IoT have non-IP/Ethernet interfaces. Consequently, they need physical translations from analog or proprietary system interfaces to enable the data to be consumed and analyzed. And that can only be done adjacent to the original device generating the data, Renaud said.
This is where edge computing in IoT can help. The edge can act as a mediator between old and , adding smart functionality to those legacy assets without modern computing capabilities.
"I was talking to someone [recently] who had a machine that had been in production in an assembly line since before the Marshall ," he said. "I don't think Amazon is going to be able to pull data from that machine. An edge device has, by definition, the connectivity to the equipment that it's monitoring or controlling. And that might be legacy connectivity, it might be [controller network] bus, it might be Modbus, it might be [Recommended Standard] 232 -- something other than Wi-Fi and Ethernet that is easily ingestible by a cloud."