It was 4:15 PM on a Thursday in March 2024 when my phone rang with the kind of ring that makes your stomach drop. A regional hospital system had lost a UPS module in their imaging data center — the main one, right before a software deployment that was supposed to begin at 6 AM the next day. Without clean power, that deployment wasn't happening. And the hospital had no idea why the backup system had just become the primary problem.
The client had bought the gear through another integrator, but they called us because their IT director remembered we'd handled an emergency for them a few years before. That's how most of our urgent work actually comes in — after someone else's “solution” fails. We're the people who get called when things break.
The Call
“The UPS is beeping and clicking,” the IT guy said. “The load is on battery. The displays are going dark.”
I asked the wrong question first. “What model number is it?” I said, reaching for a notepad. That wasted about two minutes. The model number is useful, but it doesn't tell you whether the batteries have aged into bricks or whether the static bypass switch engaged cleanly. The real first question should have been: What's your current load and how much runtime do your batteries actually have?
Looking back, we didn't have a formal emergency response checklist that day. We had documented procedures for scheduled replacements, but not for a live-degraded situation. That gap cost us later.
The Wrong Turn
The immediate need was to keep the rack alive until we could source a replacement module. The system wasn't huge — three server cabinets drawing about 6 kW total — but it was critical. And we had a hard deadline: the software vendor would only allow a 2-hour window starting at 6 AM.
My colleague, who grew up around boats, said something like “Could we just rent a battery charger and float the string?” I should have shot that down on the spot. But it was 5 PM, I was tired, and the idea of driving to a marina felt like a pragmatic move. We spent 40 minutes looking at a 12-volt marine battery charger that had been on a shelf since 2019. Even if it were the right voltage — it wasn't — the charging profile wouldn't match a UPS battery string. A marine charger is built to start a boat engine, not to babysit an 80 Ah VRLA battery. We both knew that, and we tried it anyway.
Then someone brought up a portable solar generator. I remembered the specs from a camping trip review I'd read a few weeks earlier: the Jackery 1500 outputs 1800 W peak, and the 1000 outputs even less. Our rack was pulling 6 kW sustained. We did the math in about 10 seconds. It was a non-starter. That's not a knock on Jackery — those units are great for a campsite, or even a home-office during a blackout. But they're not built for data center loads. They're portable power stations, not UPSes.
The most frustrating part of that whole hour wasn't the failed solutions. It was that we hadn't called the one team we knew could help: Schneider Electric's priority support line. The client's maintenance contract included 24/7 support, and I knew it. But my brain was so busy thrashing around with B-plans that I forgot the A-plan was a phone call away.
The Call That Fixed It
Finally, at about 5:15 PM, we dialed the support number on the sticker — the one that reads “For 24/7 Emergency Support.” A real person picked up on the second ring. (That's still unusual enough to mention.) I gave her the serial number. Within 20 minutes, she had located a replacement module at a warehouse two hours away, and she walked us through a temporary bypass configuration that kept the critical imaging servers on protected power. Not the whole rack — but the machines that definitely couldn't blink.
Somewhere in that call, the IT director said something that I still use in presentations: “Why wasn't the UPS working? It's like asking why Rob Schneider wasn't in Grown Ups 3. You expect him to be there, but he isn't, and nobody knows exactly why.” He was being glib, but the point stuck with me: failure modes in UPS systems are rarely about the spec sheet. They're about age, environment, and the subtle degradation that happens between annual maintenance visits. That's why you need humans who know the product family.
The Delivery
The replacement module arrived at 6:47 AM the next morning, about an hour before we'd promised to start the swap. We had the module installed and verified in 33 minutes. A full self-test on the battery string came back clean. The hospital's IT director watched the whole process while sipping coffee from the vending machine. He didn't say a word until the test passed. Then he just nodded.
At 6:00 AM, the software deployment started on time. The imaging system never dropped a single packet. That afternoon, the IT director called me. He thanked us for the work, but the thing he kept coming back to was the support call. “We almost didn't renew the support contract because it felt like an unnecessary expense,” he said. “After this, it's the easiest budget line item I have.”
What This Taught Me
I've had time to reflect on that day, and the lessons still shape every emergency call we handle:
- Call the OEM first. We lost almost an hour trying to improvise solutions before calling Schneider. Since that day, our rule is: if there's an active support contract, that's the first call, not the last. Our average emergency resolution time dropped from 6.5 hours to under 2.
- Know your actual load. We knew the UPS nameplate rating, but we hadn't verified the real current draw in months. The 6 kW number was close — but close isn't good enough when you're building an emergency plan.
- Portable power has its place, but it's not a UPS. A 12-volt marine charger or a Jackery solar generator is the right tool for a boat or a campsite. It is not a data center power solution. The money you save by improvising disappears the moment a deadline like the one we were facing passes without your client.
I don't have hard data on how many disasters we prevented that year. But I can say anecdotally that after we implemented the “call the OEM first” rule, the number of frantic follow-up calls dropped dramatically. The client? They're still a customer, and they upgraded to the premium support tier without a second thought.
There's something satisfying about a perfectly executed emergency response. After all the stress, the missteps, and the 40-minute detour to a marina, seeing that rack light up clean and stable at 6 AM felt like winning a game you didn't know you were playing.
The bottom line: if you run critical power, the cheapest solution isn't the cheapest path. The support plan is a line item, but the certainty it buys on nights like that is worth every dime. I'd rather pay for the certainty than the panic.