Back in Feb 2024, I was sitting in my cubicle—actually, I was standing at my desk because I'd just spilled coffee on my chair—when the lights flickered. Not a full brownout, just a flicker. Long enough for the server in the back to reboot. Long enough for the VP of Operations to call me and ask, in that tone, 'What's our plan for this?'
I manage purchasing for a 120-person electronics distribution firm. My budget this year was ~$150k across all non-inventory buys. That covers everything from office supplies to the specialty equipment our field techs need. I report to both Ops (for logistics) and Finance (for the receipts). It's a balancing act.
When I took over this role in 2020—right when everyone was scrambling—I inherited a mess of vendor relationships and a single, ancient UPS that beeped mournfully anytime someone used the microwave. The VP's question forced me to confront something I'd been putting off: we needed a real power protection strategy. And I had to figure it out without blowing my budget or making a decision that would fail us when we truly needed it.
Look, I'm not an electrical engineer. I'm the person who orders the chairs and the paper and the toner. But our company relies on a small server rack that handles our internal ERP system and a bunch of developer workstations that our software engineers use to write code. The production database? That's in a colo facility. But our development and testing environment is on-site. Every time we had a flicker, someone lost an hour of work. On a bad day, a week of changes got corrupted.
The cost wasn't just aggravation. The VP of Engineering estimated each full crash cost us about $15,000 in lost developer time and rework. And we were having about 4-5 of those a year. That's somewhere between $60k and $75k a year—just from losing unsaved work and dealing with file system corruption on reboots. That math got Finance's attention real fast.
So here was my brief: I needed to protect our critical path equipment (the server, a few key switches, and a couple of the most active developer workstations) from outages long enough for a safe shutdown. That meant a UPS. I also needed to think about what happened if the outage was longer than the UPS battery could handle. That's where the question of a generator came in.
I spent about three weeks on this. My process was basically: research online, call vendors, talk to our IT guy (who had opinions), and then worry about the cost.
When I started looking at commercial generator prices, I nearly had a heart attack. I'm talking about a basic standby unit that could handle our small server load and a few lights—about 20kW. The quotes I got, including installation, were in the $15k to $25k range. For a natural gas unit. That's a lot for my budget—roughly 15% of my entire annual spend. I had to justify every dollar.
Then there was the standby generator maintenance question. The vendor sales guy said 'low maintenance,' but I knew better. I asked for specifics. Turns out 'low' means quarterly checks, annual oil changes, battery replacements every 3-5 years, and a load test at least once a year. That's easily $500-$1,000 a year in upkeep. And you have to actually do it. A generator that hasn't been run in two years is a paperweight that cost you $20,000.
Here's the thing: I had a similar experience three years ago. In 2021, I pushed for a cheap generator quote—$5k for a portable unit. Thought I was being smart. We bought it. It sat in the warehouse for a year. When we finally had a multi-hour outage, we wheeled it out. And it didn't start. The carburetor was gummed up. We hadn't run it, hadn't maintained it. I still kick myself for that. It was a lesson in buying false reassurance. (Note to self: never buy something that requires maintenance you don't plan to do.)
So the generator felt like a big, expensive, high-maintenance question mark for our situation. A proper one was $20k+. A cheap one was useless. I had to solve the more immediate problem first: the flickers and short outages.
I did look at where to buy a portable generator as a stopgap. Big box stores, online retailers, local equipment dealers. Prices for a 7-10kW portable unit ranged from $500 to $2,000. It's tempting. I get it. For someone with a home workshop or a small retail shop, a portable generator you can roll out for a storm is a fine solution. But for a business with a server? You have to know how to hook it up safely (transfer switch is mandatory) and you have to commit to running it monthly. Most businesses don't. They buy it, stick it in a corner, and forget it.
That's the pattern I see. The numbers said we needed a generator for the multi-hour scenario. My gut said we needed to solve the flickers first—the 90% scenario that was costing us $60k/year—and have a plan for the 10% scenario. I went with my gut.
This is where my research on UPS units really kicked in. I needed something reliable, rack-mountable (to fit in our server rack), and big enough to give us 15-20 minutes of runtime for the critical equipment—enough for a graceful shutdown.
I looked at eaton rack-mount UPS reviews extensively. The Eaton 5PX and 9PX series came up a lot. What I liked was the network management card support. I'm not an expert on power, but if a UPS can email me or our IT guy when there's a problem, that's gold.
We ended up buying an Eaton 5PX1500RTN for the server rack. It's online double-conversion, which means it's always on, always conditioning the power, not just switching to battery when the power goes out. For a server handling development work, that clean power helps with stability even when the lights don't flicker.
We also bought a few smaller Eaton 5S1500LC units for the key developer workstations—the ones whose work gets corrupted. That was a separate budget line item, about $350 each.
Total spend for the UPS solution (main unit plus network card plus smaller units) was around $2,200. Plus installation labor (some cable management, configuring the shutdown software) was another $300 from our IT guy's hourlies.
The Eaton 1500 UPS units (we got the 5PX and 5S) were the sweet spot for our needs. They were rack-mountable, had enough runtime, and I verified they had proper invoicing and support from the vendor (remember that handwritten receipt lesson from 2021!).
Part of me wanted to consolidate to one vendor for simplicity—just buy the whole generator + UPS from the same place. Another part knew that redundancy saved us during that supply chain crisis. I compromised: the UPS came from one distributor (Eaton partner), the generator plan came from a separate electrical contractor.
I didn't completely abandon the generator idea. The VP of Ops asked about it, and I had to give an honest answer. I said: 'For what you need it for—keeping the server and a few workstations alive for 4-6 hours during a multi-hour outage—a proper standby unit is $20k plus $1k/year maintenance. That's $30k over 5 years for maybe one or two uses.'
I proposed a different approach: budget $10k of our contingency fund for a 'rental or emergency purchase' of a portable generator if we ever have a multi-hour outage. That's a fraction of the 'buy now and maintain forever' cost. The finance director liked this. It's a risk-based approach.
We agreed. We'd buy the Eaton UPS now to solve the flicker problem, and we have a plan for a generator if we ever truly need it. It's not perfect. The alternative was missing a $15,000 event when the next big flicker hit, which we avoided.
This whole experience solidified something for me. When it comes to critical business infrastructure—the stuff that, if it fails, costs you real money—you do not cheap out on the reliability of your solution. You pay for the certainty that your workstations and servers will have a clean, managed shutdown during the most common failures.
Three things I'd tell anyone facing a similar decision:
I still have mixed feelings about the generator decision. On one hand, we have a plan and saved money. On the other, if a 3-day outage happens, my phone will ring, and I'll have to explain that 'the plan' is to scramble for a rental. (Part of me wants the generator just for peace of mind. But Finance is happy with the saved money.)
As of January 2025, our Eaton UPS has saved us from at least six flickers in the past ten months. The server hasn't had an unexpected reboot since. That's probably saved us $40k in lost developer time. The VP of Engineering sent me a nice email about it—first time I've gotten kudos for an infrastructure purchase. (Note to self: capture that email for next year's budget justification.)