Culture Eats Technology for Breakfast (with a Side of SD-WAN)
Phil Howard's Dissecting Popular IT Nerds, with Doug Edmunds of Summit Design and Engineering. A real SD-WAN deployment across North Carolina, why MPLS is not dead but is not the default either, and the line the episode is named after.
31 October 2025 · 3 min read
Dissecting Popular IT Nerds, on Telecom Radio One, now part of You've Been Heard. Episode 25, published 31 October 2025 and recorded in the WAN Dynamics years. Host: Phil Howard. With Doug Edmunds, Director of IT at Summit Design and Engineering Services.
Phil is the most bearded man in telecom and runs a peer-to-peer show for IT leaders. This one paired me with a customer, Doug Edmunds, whose company we had just moved onto SD-WAN, so it is half case study and half the business-acumen conversation Phil likes to steer toward. Doug and I both came to IT from playing in bands, which explains a lot about both of us.
The problem Summit had
A fast-growing engineering firm spread across North Carolina and into Virginia, with an aging phone system bridged between two offices, field offices of twenty people with no phones at all, and site-to-site VPNs through the firewalls that made opening large CAD files painful. Nobody had been doing strategic planning; it was plug-the-dike every day. Doug's phrase was that he wanted to pave the highways first. We sized VeloCloud edges per site, decided where high availability was worth it and where a satellite office could do without, upgraded the circuits to 50 and 100 meg fiber, and layered the overlay on top with voice prioritized.
MPLS is not dead, but
Phil's position was that MPLS for mid-market site-to-site is dying because it is too expensive and too slow. I added the third reason: the applications do not live on a private network anymore. They are in Azure, AWS or Office 365, and a wholly private WAN does less for you than it used to. That said, I told Phil the same thing I told call centers with an on-prem Call Manager or Mitel system: do not go all-internet. Keep one MPLS circuit at the key sites, pair it with internet, and let SD-WAN pick the path.
Picking a flavour
Sixty-odd vendors claimed SD-WAN at the time and roughly 30% of domestic companies were exploring it, so I told listeners to do their homework, because a lot of what was on the market was WAN failover with a controller, which is glorified load balancing. We had chosen VeloCloud four years earlier, and the early days were rough, but it handled voice better than anything else, with forward error correction and jitter buffering that steer around problems in real time. What Doug valued most was not the failover. It was visibility. With a thin bench and no dedicated network team, he could open the orchestrator, see how the network was behaving and rule things out before chasing a ticket. He also learned enough of the business policy model to fix his own VoIP routing after watching our engineer do it once. I told the factory story again: 60% of one site's bandwidth was YouTube because everyone built a playlist at 8am.
The line
Phil asked each of us for the single trait behind our success. Doug's was communication, upstream and downstream, and his go-to opener with frustrated users is "I hate technology," which takes the air out of the room. Mine was that most business problems people try to solve with technology are actually culture, process or communication problems, and you have to step back and ask what problem you are solving before you decide whether technology is even part of the answer. Culture eats technology for breakfast. I have watched organizations spend millions on a platform their own silos would never let succeed. Phil's takeaway list for the episode: culture eats technology for breakfast, I hate technology, end users are our customers, and communication. Also, there is an RFC for IP over avian carrier, and Phil was briefly considering a career in it.
