You’re on a videoconference call strategizing about the next product launch when your screen freezes. Your collaborators continue working together, unaware that you are no longer able to participate. Eventually, your connection resumes, but you begin to notice delays between when you speak and when participants hear your voice. Or maybe you’re working from home, trying to send a large file to a colleague without success. You try again later, and it finally goes through.
Availability, latency, and reliability issues like these have always been challenging, but they have become much more pervasive today. In the current environment, workers collaborate from a variety of locations, customer service relies more on video, and businesses require more bandwidth and five-nines of availability for real-time applications and communications.
A big part of the problem is that most networks were not designed for real-time traffic. Many, in fact, still have remnants of the traditional hub-and-spoke campus infrastructure, which isn’t very compatible with the cloud. And then you must consider the way cloud services work; they typically don’t provide any real-time guarantees, and performance can vary because compute and network resources are shared.
Another issue is the sheer number of applications running in real time. The issue has grown steadily since the introduction of 3G, which spurred development of a few voice applications, followed by explosive growth in real-time applications — think Twitter and FaceTime —when 4G came alone. 5G is pushing the envelope with more bandwidth-hungry video apps and even augmented reality
Much of the latency and availability issues occur during the “last mile,” or that 10 to 50 or so meters between the carrier’s edge and the home or business. Even with excellent connectivity at your business sites or home offices, the quality of that last-mile connection depends on how good your WiFi infrastructure, cable connection, or 4G/5G service is.
“Before workers went home, businesses would buy their own private network connections, or buy connections to Zoom, [Cisco] Webex, RingCentral, and others, but they have less control today,” Zeus Kerravala, founder and principal analyst with ZK Research, said. Today, about 90% of connectivity issues crop up in the last mile, he estimated.
Alagu Periyannan, vice president for advanced collaboration technologies at Verizon, agreed.
“Often, it depends where you are sitting and where your WiFi routers are,” he said. “It can even happen in businesses where WiFi infrastructure hasn’t been deployed correctly. Certain conference rooms might get a great signal where others don’t. It’s part of the last-mile problem.”
The responsibility for improving availability, reliability, and latency lies mainly with bandwidth providers, collaboration technology vendors, and network resellers, but some of it also falls to businesses themselves.
Companies that haven’t moved off of their closed networks and traditional VPNs should take that action first, recommended Curtis Peterson, senior vice president of cloud operations at RingCentral.
Relying on traditional VPN technology during the switch to remote work didn’t go very well for some companies. Many ran out of simultaneous licenses, stretching the technology beyond its limits, with lots of bandwidth constraints at concentration points. That’s likely to continue as people return to a hybrid work environment. Instead, companies should implement some type of on-campus wireless strategy like WiFi version 6, private LTE, or a private 5G network, Peterson advised.
Moving to some type of software-defined network is another important step. An SD-WAN, for example, can increase availability and provide insights about the type of traffic that isn’t typically available with standard routers and firewalls. Software-defined technology also allows organizations to run multiple connections, monitor the network for performance, and guide traffic toward the best path. It’s a good step toward solving the middle-mile problem — the segment between the core network and last-mile providers. The middle mile can range from tens to thousands of miles.
Most importantly, companies need to make sure they have the information they need to ensure that traffic gets where it needs to be, on time and intact. That means knowing where and how bandwidth is being consumed.
“If your CEO is trying to have a conversation in his car and isn’t using a communications provider that can reduce bandwidth needs as he moves through 4G or 3G spots, then you have failed,” Peterson noted. “It doesn’t matter that you knew that the CPU of the router at the main building was below your target.”
That requires creating a mobile-friendly and mobile-first environment and making sure you have the analytics you need to monitor your traffic, usage patterns, and connections. The right analytics can make all the difference in figuring out where your problem spots are, which is the first step to fixing them.
Technology with Vision
As the use of real-time communications skyrockets, vendors are continually evolving their approaches to solving for these new traffic demands.
For example, managed SD-WAN service provider Masergy (set to be acquired by Comcast Business) recently announced Performance Edge, which the company says uses proprietary network architecture and industry standard routing algorithms to bring Ethernet-like performance to a single broadband connection. Performance Edge
is available with Masergy’s managed SD-WAN and secure access service edge (SASE) offerings.
Cisco has announced partnerships with colocation and public cloud providers to extend the reach of its SD-WAN offerings. These partnerships help Cisco provide more optimized routing directly to the workload and the ability to take advantage of the massive, high-performance backbone networks connecting each of these providers’ data centers.
Building networks around real-time communications starts with a software-defined and highly programmable network fabric that allows network engineers to make fast updates as processing demands change, Jeff Scheaffer, Cisco’s vice president of management and security for software-defined network transformation, said. In addition, network engineers need visibility across network and application performance so they can ensure quality application experiences for users and to fix problems quickly,” he added.
Other vendors are trying to solve the problem with different approaches. Videoconferencing services provider BlueJeans, co-founded by Periyannan and now owned by Verizon, focuses on solving the last-mile issue with a focus on wireless connectivity.
“If you’re on the East Coast and I’m on the West Coast, the round-trip latency is about 80 milliseconds from a speed-of-light perspective,” Periyannan said. “But the Wi-Fi in your house or office building can add anywhere from 100 to 300 milliseconds because of the nature of the unlicensed spectrum you're using.”
To solve that problem, BlueJeans has developed techniques
to adjust bandwidth as requirements change, along with packet loss techniques to fine-tune that last 50 meters. This allows the technology to adapt to wherever users are connecting, at any time.
This will become even more important over time, especially as 5G becomes more ubiquitous. Verizon is working with Amazon to run some of the infrastructure right at the edge of the network. The goal is to reduce latency while improving consistency and availability, Periyannan says.
Subspace, a company dedicated to building a global networking platform optimized for individual traffic streams, takes a different approach
to the last-mile problem. The Subspace platform constantly monitors Internet traffic, using its technology to find the shortest and least bottlenecked path at the time it is needed.
So, if you are on a Zoom call with someone across the globe, the Zoom traffic will run across the Zoom network between Zoom nodes. Then comes that problematic last mile, and that’s where Subspace can help optimize the connection. It does this by building out hundreds — eventually thousands — of purpose-built points of presence in data centers worldwide. Each PoP, which has its own computer power, can handle millions of servers.
“We focus on making sure that the traffic is smooth, and where connections are bad, we’ll actually go out and buy our own wavelength,” Mo Nezarati, Subspace’s president for voice, explained. “Essentially, we are leveraging the public Internet, private cloud paths, and even some of our own paths we’ve purchased, with the goal of getting traffic from point A to B faster. Our goal is to be milliseconds away from the end user.”
No matter which route a company chooses, the key is preparing for more of same — more applications running in real time, more 5G, more user demand, and more advanced technologies like augmented reality. That means both looking inward — assessing your needs honestly — and choosing communications and network vendors that can meet those needs.