Ethernet Switches: Why Time Matters

29 March 2016
MQ-9 Reaper

It’s no coincidence that our blog is called The Connected Battlefield. Communications networks are at the heart of modern military operations. But there’s more to the name than that. At Abaco, networking—and especially rugged networking—is fundamental to what we do. 

What that means is that we spend a great deal of time understanding our customers’ problems—and helping solve them. Here’s a recent example.

Ethernet technology now dominates military networks, and at the heart of those networks are Ethernet switches. There are fundamentally two kinds of Ethernet switches: unmanaged, and managed. Unmanaged switches are simple “plug-n-play” units that enable devices to talk to each other. Managed switches, on the other hand, are significantly more complex because they provide the user with far more flexibility to control those conversations. 

Managed switches enable you, for example, to use protocols like SNMP to monitor the network and the status of the devices on it. They allow the creation of VLANs to segregate/segment traffic. They allow prioritization of data flows—perhaps classifying time-critical traffic by Quality of Service (QoS). They allow you to design-in redundancy to mitigate the effects of partial network failure. They add a level of the network security—maybe checking authorization. And, of course, much more…

But—and there’s always a “but,” right?—this extra capability comes at a cost. Yes, a managed switch will always be more expensive—but, perhaps more importantly, all that additional functionality and performance means that a managed switch takes longer to become fully operational once it’s powered on.

Not an issue

In many applications and environments, that’s not an issue—even when boot time is measured in minutes, as it is on some managed switches. In an office environment, for example, a switch is likely to be powered on only once—and then left to run.

But what about environments where, for example, the requirement is to be able to turn the switch off when it’s not in use to save power or minimize heat dissipation? That’s the case in many military applications. Those mission-critical applications can’t wait several minutes for the switch to become fully functional—or even the 30 seconds typical of Abaco managed switches. When a switch is responsible for getting sensor-derived data (from cameras, for example) to a military vehicle’s driver or pilot, and when that sensor-derived data is all he or she has to go on—seconds count.

We had a customer who really needed all the capabilities of a managed switch—but he also really needed a boot time much closer to that of an unmanaged switch. At Abaco, that’s the kind of challenge we relish.

Did we rise to the challenge? Yes, we did. Interested to know how? Check out our new white paper—“Ethernet switches: why time matters”—here.

And: if you’ve got a tough networking challenge you think we may be able to help with—please get in touch. Networking is what we do. 

 

John Thomson

John Thomson has been working on software for over thirty years, having done a degree in Computing Science in the early 80s at Glasgow University. His focus has always been networking, in particular Local Area Networks, including years working on international standards for protocols - back in the early days of the Open Systems Interconnection (OSI). Having worked for a number of multinational companies (both large and small) he has been with Abaco (or its predecessors) since 2000, and leads the team of software people working on OpenWare – our Ethernet switch management suite. The OpenWare team is based in Edinburgh and California.

 

He and his wife took a career break for five years in the early 1990s, and were heavily involved in relief and development work (with Unicef and others) in Ulaanbaatar, Mongolia. He still claims to remember enough of the Mongolian language to ask for some very interesting foodstuffs!