By the year 2025, over 200 million edge nodes will be deployed worldwide. But what will all this additional spending on edge computing amount to? For distributed enterprises, those with many locations or those with a far reach, edge computing can reduce latency, lower the load on central data centers, and increase security. So how do you implement edge computing, and why should you?
What is edge computing?
Edge computing is actually a fairly literal name. Companies have centralized data centers where they store large amounts of information, but to get that information to customers and employees far away from the central data center, they need to push some of that computing power to the edge of the network – and expand the concept of enterprise networking to include the needs of all those remote devices. With these edge nodes, customers and employees can get information faster than they could if they needed to access the central data center. The OpenStack diagram below shows how a network that features edge computing might be set up.
Edge computing benefits
Reduced workloads on servers
In traditional data center models, whether on premises or cloud-based, when remote applications need data, they send a signal all the way to the main data center and then have to wait for it to come back. While information travels quickly over the internet, even minor delays can add up over the course of a day or week, and the more relays there are, the higher the chance that a request hits a delay. In critical use cases, say autonomous cars or energy sensors, latency and reliability issues can have serious consequences.
Edge computing, on the other hand, utilizes local resources to process data, reduce latency and allow applications to perform actions based on predetermined triggers. If a company needs to update a trigger, then it can send updates via the data center.
For example, a manufacturer may use IoT devices in their production environment to automate certain parts of the assembly line. Rather than having the IoT device send a signal to the central data center every time a new product triggers it and the data center sends instructions back, the manufacturer can keep data processing nearby via an edge caching server or load instructions onto the IoT device itself and have it respond to the trigger on its own.
In the case of a self-driving car or critical infrastructure monitoring, real-time data processing becomes essential, requiring data processing on the device itself and rapid communication via local infrastructure, making 5G a critical driver of edge computing and the Internet of Things.
With edge devices able to process data and communicate, the networking model becomes something akin to the peer-to-peer networks of the early 2000s, communicating with data centers when necessary.
Also read: Where AI Can Improve the Supply Chain
Edge computing can also help companies become more flexible. Businesses can set up local edge data centers to extend their reach into new markets rather than building entirely new offices. Additionally, because IoT devices are always on with edge computing, they can provide businesses with an unprecedented amount of information. All this data leads to greater insights for the business, allowing them to easily pivot or make adjustments if something isn’t working as it should.
Edge computing is also key to enabling remote work, especially as 5G infrastructure becomes more mainstream. Currently, edge nodes aren’t widely used enough to reach all residential areas, but once 5G is widely available, businesses could combine it with their edge networks and extend them to their customers and remote employees.
Because edge computing is less centralized, companies that use it are less vulnerable to distributed denial of service (DDoS) attacks. With so many edge data centers and servers working simultaneously, it’s difficult for attackers to manipulate enough traffic to shut them all down. While there are more points to secure versus having a single data center, the distributed nature of the edge removes any single point of failure. Edge computing also makes it easier to quarantine affected devices and remediate threats without shutting down the whole network.
Along with spreading out the attack surface so it’s more difficult to make a widespread attack, edge computing also transfers smaller amounts of data at once. While transferring data would normally be vulnerable to attack, the smaller amounts make it a less appealing target to malicious actors. Plus, devices only communicate with the data center when necessary, like for updates or data refreshes. This way, even if a device is breached, the amount of data that is compromised will be much smaller.
Reduced workloads on servers
The average home in the United States contained 10 internet connected devices in 2020, and that doesn’t even include business devices. Each of a company’s devices generates an insane amount of data which could easily overload servers if left unchecked – a risk that will increase with the huge data demands of artificial intelligence (AI) and remote devices. However, edge computing spreads out the processing of this data into multiple locations to reduce the workload on central servers.
Most businesses worry about DDoS attacks, but even legitimate traffic could be a danger to centralized servers in the right circumstances. For example, a 24-hour gym franchise might use IoT scanners to control access into the facility. If all of those devices were sending and receiving signals from the central system, that would be a pretty inefficient way to run a network and would also create a single point of failure that could shut everything down. Instead, the sensor just reacts to the trigger and allows the approved individual inside, updated by the central data center only as information changes.
The importance of AI in edge computing
AI is going to become a critical part of edge computing as it becomes more mainstream. Self-driving cars are a great example for this. When self-driving cars become a reality, they’ll need the ability to take in and process information in real-time. To do this, the cars will include AI to perceive the world around them, make inferences based on the data they take in (in this case, traffic signals and other cars on the road), and then act based on that information.
Edge computing is key to this because real-time processing wouldn’t be possible if the cars had to send all of the information they take in off to the central data center and then wait for it to come back. Instead, the car itself would become an edge node, able to process all of the data within its own CPUs. Peter Levine of venture capital firm Andreessen Horowitz likens this model to the peer-to-peer (P2P) networking of the early 2000s, where devices communicate with each other and contact data centers only as needed.
Implementing edge computing
Edge computing isn’t necessary for every company, but those with remote data processing needs could gain a lot from it. If an organization decides to move forward with edge computing, it can either build the infrastructure itself or contract with a third-party vendor to provide the necessary hardware and software.
In some cases, say with critical infrastructure, businesses may want tighter control over their edge computing capabilities. If so, they may decide on a do-it-yourself (DIY) model. The business would need to purchase edge computing hardware from vendors like Dell or HP according to their network needs and then add the right management software. Then, the company’s IT department would need to set everything up and ensure the devices and software are connected and compatible. While this offers more customization and flexibility, it’s a lot of work that smaller businesses may not want to undertake.
If the cost is too high for companies to set up their edge computing network on their own, they may opt instead to contract with third-party providers. For this option, the vendor would install their own equipment and software, and the company would pay a flat monthly rate. The vendor would handle all of the updates and maintenance, reducing the organization’s need for dedicated IT staff. This option requires fewer resources from the purchasing company, but it doesn’t have the same level of control or customization.
Intelligence in devices themselves will also be critical to edge computing, so businesses will want to evaluate their options carefully.
Is edge computing the right move for your business?
Edge computing is a great option for businesses with many locations or that need to reduce latency for customers far away from their headquarters. Streaming services, large franchises, and companies that manage their own supply chain could benefit from edge computing. You should take stock of your current computing requirements and note any latency that you’re seeing. If you’re seeing time loss or your servers are getting close to overloading regularly, it may be time to consider edge computing.