I was working for a company with almost 3,000 employees and up to 5,000 guests on-site at any given time. When I walked in, the internet was a mess.
We had 5 to 7 separate AT&T circuits terminating into the building. All combined: about 2 gigabits. We were paying $15,000 a month for bandwidth that couldn’t keep up during a busy day.
If anyone’s ever worked with AT&T on an enterprise upgrade, you know the drill. Two to four weeks to schedule the meeting. 10, 12 people in the room, all “helping.” The new design was more expensive, more complicated, and somehow still not enough. I needed 4, maybe 5 gigs to handle peak events safely. AT&T’s timeline measured in quarters, not weeks.
I called the usual suspects, Comcast, Verizon,for a backup ISP. All about the same.
Then I got a call from a rep at Hotwire Communications. Full plug, they deserve it after what happened next.
The site survey that changed everything
Hotwire had just extended a fiber loop near our property. Were we interested? Absolutely. I was looking for a secondary ISP. We scheduled a site survey.
The tech walked the property. Standard install. Then, as he toured the rest of the building, he said something offhand:
“You guys are really close to our other fiber loop.”
I didn’t think much of it at the time. We finished the survey, talked secondary ISP pricing, everything was normal.
When the proposal came back, the tech mentioned that the second loop was across the street from our other demarc location. Immediate question: does that second loop go to a different data center than the first?
Yes.
My brain kicked in.
Two fiber loops. Two different data centers. And I happened to know, from dealing with AT&T’s legacy build,that there was 4-inch conduit connecting our two demarc locations underground. The previous team had installed it years before I got there.
I didn’t need a secondary ISP for redundancy. I needed two paths to two different data centers from the same ISP, so I could do real routing-level redundancy with my own IP addresses.
I called the sales rep back. “Am I reading this right? Two different loops, two different data centers, you can provide service to both?”
Yes.
The carrier-grade wake-up call
We started talking pricing. She said that at 4 or 5 individual 1-gig pipes, it was actually cheaper to just buy one 10-gig pipe.
I said great. Not fully understanding what a 10-gig pipe actually means.
Here’s what it means: to terminate 10 gigs at each demarc, Hotwire needed to install about $100,000 worth of Cisco equipment per location. That was my introduction to carrier-grade pricing. The rule of thumb: every zero you add to your WAN speed, add a zero to the equipment price. A 10-gig ISP handoff? $100,000 per location. I needed two.
$200,000 in infrastructure costs. On top of roughly $50,000 in firewall upgrades on my side. I couldn’t sell it.
The deal
We went back and forth a few times. Then the rep came back with a counteroffer.
Hotwire’s two fiber loops didn’t talk to each other yet. They were paying a third party to connect their data centers. If they could run fiber through my building, using that 4-inch conduit connecting my two demarcs,my property would become part of their backbone.
The offer: they’d eat the $200,000 in infrastructure costs if we gave them 24/7 access to the demarc locations.
I said: “So you’re not going to charge me $200,000 for equipment, and you don’t want me to charge you rent for hosting your backbone in my building.”
Basically, yes.
I ran the numbers. $50,000 for new firewalls and 10-gig WAN upgrades on our side. Doable. I got budget approved, and we built it.
10-gig pipe on both ends. Redundant router configs on our side. Redundant infrastructure on theirs. Cross-connected at both demarcs. Two paths to two different data centers through the same ISP, broadcasting our own IP addresses.
The numbers
A month or two later, everything was running beautifully. More bandwidth than we could use. Then I started doing the math on what we’d actually accomplished.
When AT&T’s contract expired, their bill climbed to over $20,000 a month. For 2 gigs across 7 circuits. My new Hotwire service, 10 gigs of dedicated, redundant bandwidth across two data centers,cost around $7,000 a month.
Let me say that again. 5 times the bandwidth. Two independent paths. Full redundancy. Less than half the cost.
And it didn’t stop there. I didn’t technically have 10 gigs of internet. I had 20, ten on each side. Around the same time, Hotwire was published as the fastest ISP in the United States. I asked them: had they ever installed 20 gigs anywhere else? Never.
I went to our marketing team. “We now have the largest internet deployment on the fastest network on the East Coast.” We hedged with “East Coast” because there’s always Las Vegas.
Marketing never sent the press release. I don’t think I sold it well enough. But anyone reading this probably gets it.
Here’s what I took away: I didn’t set out to build a 20-gig redundant internet deployment or save $13,000 a month on telecom. I set out to fix a bandwidth problem that AT&T couldn’t solve on any reasonable timeline. The opportunity came from showing up to a site survey, asking the right questions, and recognizing that what my ISP needed was something I already had.
The $200,000 in free infrastructure was just what happened when both sides of the table had something the other one wanted.