Jagged Edge: Four Myths About the Edge of the Network
The data center is changing, and with it the physical infrastructure. Today, the traditional enterprise model is just one potential architecture among many emerging designs. The distributed data center, with computing at the edge, is increasingly common – but the edge of the network is no less important, and infrastructure in those spaces no less critical, than traditional environments. In fact, these edge sites often present their own unique challenges.
The data center is changing, and with it the physical infrastructure. Today, the traditional enterprise model is just one potential architecture among many emerging designs. The distributed data center, with computing at the edge, is increasingly common – but the edge of the network is no less important, and infrastructure in those spaces no less critical, than traditional environments. In fact, these edge sites often present their own unique challenges.
Unfortunately, many organizations have been slow to grasp the nuances and increasing criticality of edge computing. This lack of understanding is compromising network performance. Solving the problem starts with identifying some of the myths about edge computing and accepting these sites and the unique challenges they present to network administrators and data center managers.
Let’s look at a few of those myths, starting with the most fundamental.
Myth No. 1: "The edge" is just another way of saying "IT closet"
Even if the average IT manager doesn’t really believe this, some habits are hard to break. It’s understandable; many of these edge computing sites look more or less like traditional IT closets and computer rooms – single racks or potentially a row of servers with enough on-site computing power to service your average office. Other than a few extra wires leading out of the room, it would be hard to tell anything else is happening in these spaces.
Ah, but those wires change everything. Those servers are part of the larger network, one module in a distributed data center architecture that expands far beyond any four walls. And the edge is only increasing in importance as our environment becomes more intelligent and connected. Sensors and software tracking personal finance changes, unusual privacy-related activity, traffic patterns, shopping trends – this is the Internet of Things and beyond – are capable of rapid data analysis and triggering responses when supported by edge computing.
In short, the edge of the network is far more valuable – and outages or interruptions far more damaging – than any issues in your old office IT closet.
Myth No. 2: You can cut corners with your UPS
It’s not that anyone ignores power protection at the edge, it’s just that all too often it isn’t sufficient. We’ve seen everything from simple, outdated UPS systems to off-the-shelf power strips deployed in the interest of "power protection" in edge spaces. Understanding what we do about the importance of the edge (see Myth No. 1), this is unacceptable. These edge systems require clean, consistent power, and that starts with a robust UPS system.
Even the location of edge computing equipment should be considered. For example: We evaluated a site that consistently experienced unusually short UPS battery life. What we found was the IT equipment was located too close to the elevators, meaning the UPS was forced to handle repeated power sags that happened every time the elevator was operated. Those sags drained the batteries, and when they really were needed, they couldn’t deliver the backup power the site needed.
This is an increasingly common problem. As more and more devices join the grid, there are more sags and spikes in electricity use. There are good UPS systems designed for small spaces at the edge that monitor and manage these issues to ensure the computing assets receive clean power and are protected in the event of an outage.
Myth No. 3: If devices are on the network, they are communicating
This one seems reasonable enough. Everything is on the network, so of course it’s communicating … right?
Not necessarily.
These edge facilities bring IT closer to end users, connecting back to data centers that act as the hub of the network. Individually, these are simple IT resources, but taken collectively, these are complex networks with considerable management challenges – challenges requiring intentional, sophisticated machine-to-machine communication.
There are plenty of technologies and software systems that make true, intelligent edge communication possible. It can be as simple as a console server, which ensures secure access to remote servers, or more sophisticated to incorporate some level of data center infrastructure management. Scalable, modular options allow organizations to build a DCIM solution that fits their needs – even if those needs are scattered to the far edges of the network.
But whatever the scale, secure remote management ensures these distributed data centers operate as holistic IT assets and not simply loosely connected collections of IT equipment.
Myth No. 4: Security starts at HQ
You’ve probably read a lot about cybersecurity over the past three years – and with good reason. Sony, JP Morgan Chase and eBay are among the international heavyweights rocked by cyberattacks in recent years. The lesson is clear: there are vulnerabilities everywhere. The 2013 Target attack originated with an HVAC vendor. Hackers can gain access through anything connected to the network – and that includes everything at the edge.
Despite increasing awareness of cybersecurity and threats, the edge remains vulnerable, often seen as an unmanageable security challenge. Some organizations have hundreds or even thousands of computing nodes at the edge of their networks. Security at those sites is a massive – but unavoidable – undertaking.
Understand this: the cost and frequency of data breaches continues to rise, despite increasing awareness of cyber threats. The Ponemon Institute Cost of Data Breach study shows a 23 percent increase in the cost of a privacy-related data security breach in the last two years, with the total cost now at $3.8 million. In addition, a Ponemon study of data center downtime commissioned by Emerson Network Power found the number of security-related downtime incidents rose from 2 percent in 2010 to 22 percent in 2015.
We may know more about the problem, but it hasn’t translated to preventing it. Increased attention to vulnerabilities in these distributed networks is essential. Today’s data center technologies are designed to prevent unauthorized access and harden the network against cyberattacks.
