Design, deliver, and run enterprise blockchain workloads quickly and easily.
All servers and systems
It’s not all that unusual for military technology to eventually trickle down to the consumer level. While it’s taken a few decades, a networking technology designed for the battlefield is now increasingly common in enterprises and in your home.
Most people who think about networks probably envision a hub-and-spoke topology. A device (say, your laptop) talks to a central hub. If the network traffic calls for it, the hub may talk to another central hub that has another bunch of devices (maybe a server) hanging off it. It’s a hierarchical structure.
Hub-and-spoke is how a phone network works, too. To be ridiculously simplistic, your phone talks to a central office. The central office either routes your call to another phone connected to that central office or it sends the call to a phone that’s linked to a different central office.
This network structure has become so commonplace that it feels like an immutable law of nature. Except that’s not necessarily the case.
As far as computing technology goes, hub-and-spoke networks arrived with Ethernet in the mid-1970s. Before that, when mainframe and minicomputers roamed the earth, computers were networked using a ring topology, not unlike the way Christmas tree lights are wired.
And now, an increasing number of wireless networks are adopting networks based on a more efficient mesh topology.
The problem with conventional hub-and-spoke Wi-Fi networks is that the hubs are dumb. They only know about the devices that are connected to them and the one hub upstream from them. Moreover, the relationship between hubs and devices is a little needy. Once a device and a hub connect to each other, they try to stay together even if a better connection becomes available. Neither the device nor the hub knows something better is out there until it's forced to look. The result can be devices and networks that benchmark fine but, in production, operate slowly because they’re not using the most efficient connections.
Mesh networks operate in a different way. Mesh hubs are aware of all other mesh hubs within radio range, which lets them manage traffic in a way conventional Wi-Fi networks can’t. Rather than creating a hierarchy of hubs, mesh networking creates a self-managing web of connectivity. As your laptop or cell phone moves around a network, the mesh network can tell if you’d be better off connected to a different hub with less congestion or a stronger signal, and it routes your traffic accordingly. When a mesh hub fails, other nodes may take over the traffic, where a hub-and-spoke network would leave devices high and dry.
A mesh network requires a much higher level of intelligence than does a classic Wi-Fi network. It also needs out-of-band signaling between the nodes, which classic Wi-Fi networks don’t require. That out-of-band signaling is a strength of mesh networks, as it allows the entire network to be aware of the network’s status and make necessary adjustments.
The downside? Because those signaling protocols are propriety to each vendor, it means gear does not interoperate. The interoperability problem is hardly news. The IEEE has been working on an interoperability standard—802.11s—since 2004. It’s not clear when the standard will come up for a vote, let alone when the standard might be adopted.
The theory and technology behind mesh networks is not new. First used in military applications—where reliability, simplicity, deployment speed, and robustness are vital—and later by first responders, mesh networks have become common in sports arenas and large public areas where flexibility and ease of maintenance are key.
About a dozen consumer-grade mesh Wi-Fi systems are on the market. They typically come in sets of two or three nodes, which the vendors claim can provide enough coverage for a house as large as 6,000 square feet. But the fact that they come in sets is a giveaway to the lock-in problem. You can’t mix, say, Linksys Velop nodes with D-Link or Google nodes, despite the fact they’re all running 802.11ac. The signaling protocols are different—and proprietary.
Similarly, there are many vendors of enterprise mesh networking gear— Aruba, a Hewlett Packard Enterprise company, being a leader in the field—with typical deployments in stadiums or other large outdoor public spaces. Metro-area networks deployed in smart cities use large deployments of mesh networking. If you use public Wi-Fi, you’re connecting to a mesh network.
The intelligence built into mesh networks helps with their setup and maintenance. Fixing dead spots with conventional Wi-Fi requires a fair amount of guesswork about the placement of range extenders or access points. When coverage is thin, a mesh network scheme can explicitly inform a device—a smartphone, for instance—and suggest where a new node would be helpful. There are certainly specialized tools to do that with conventional Wi-Fi, but mesh nodes make it easier.
One technical criticism of mesh networks is that throughput deteriorates with each hop through the mesh. In a high-bandwidth Wi-Fi environment like 802.11ac (with speeds as high as 1.3 Mbps), the throughput loss is acceptable if you can keep the number of jumps between the device and gateway to three. Conversely, the bandwidth needs of home automation in ANT Blaze, Bluetooth Low Energy (BLE), or Zigbee networks are minimal, so throughput losses don’t matter all that much.
Sonos, the connected speaker company, uses mesh networking in its products. A Sonos network uses a home Wi-Fi network, but each speaker is a node that connects to every other speaker with mesh technology. That allows one-touch setup (the intelligence happens within the mesh network) and lets customers group together speakers, which can be placed even in Wi-Fi dead spots, as the mesh network can get to places that a conventional Wi-Fi access point might not reach.
Also, ANT networks use a mesh topology. ANT, a technology owned by Garmin, is typically used for heart-rate monitors. But it can also be used for home automation. ANT nodes don’t require much in the way of power or bandwidth but can extend to hundreds of devices over a large physical footprint. A new version of ANT—ANT Blaze—has been adopted by office furniture maker Herman Miller for a line of smart furniture. ANT Blaze claims the ability to scale to 500 nodes.
For its part, BLE also now supports a mesh topology, particularly for Internet of Things applications. Similarly, Zigbee and Thread-based smart home devices set up and use mesh networks that require no human intervention.
The ill-fated One Laptop Per Child (OLPC) project was also built around mesh networking. Every machine was capable of automatically networking with every other nearly OLPC laptop, allowing students to work collaboratively.
Mesh networks show lots of advantages over older topologies, whether you’re looking for rapid deployment, robust infrastructure, or just the ability to serve bandwidth to thousands of people moving around your coverage area. More and more mesh networks are around you—at work, when you’re out and about, and when you play around with smart homes. The flexibility allowed by their topology and intelligence make them an easy choice for many types of corporate networking and home applications.
This article/content was written by the individual writer identified and does not necessarily reflect the view of Hewlett Packard Enterprise Company.