Hardware: Learn How to Do More With What You Have

Understanding hardware isn’t glamorous, but it’s the bedrock of DevOps. Get it right, and you’ll deploy more and waste less.

When Cloud Was Just a Cloud

In the beginning, when “cloud” was just that fluffy thing in the sky you stared at until it looked vaguely like a sheep, the world was powered by on-prem servers. These computing beasts dwelled in dry, air‑conditioned rooms at the back of the office, quietly humming away. They were visited only on rare occasions, usually by someone wearing glasses and a look of mild panic, mostly to swap a port on the patch panel or when something truly catastrophic happened. Don’t get me wrong, there are plenty of these on-prem devices still out there today, working non-stop, blissfully unaware that one day they’d be listed as an endangered species, right between fax machines and dial-up modems.

Blueprints Meet Reality

How do these servers end up in that mysterious back room, you ask? It all starts with a Project Manager and an idea—usually followed by terms like “make it fast” or “don’t go over the budget”. From there, you accurately calculate how much processing power you’ll need… then double it for good measure. First comes the CPU, which is almost always the default option, because picking a custom one adds weeks to the order and several progress report meetings with PM. RAM? You wing it. It’s cheap, there are plenty of slots, and you tell yourself, “I can always add more later,” ignoring the downtime that will ruin someone’s weekend. Storage —that’s where you think a little harder: how much for the OS, how much for data? Then you triple it so you can do a nice RAID 10 setup—assuming you remembered to order the storage controller, of course.

The Vendor Curveball

You now have the full specs and feel great about what’s on paper like an architect admiring blueprints. Then reality strikes when your supplier replies. The model you wanted is End of Life. Instead, you’re offered a shiny new replacement, along with components that don’t match your original specs but are “guaranteed to complement your build wonderfully”—which is vendor-speak for “good luck.” And of course, the new model comes with something extra: you guessed it, extra price tag. You forward the quote to the Project Manager with Finance in CC, explaining why we’re suddenly over budget, and pray it only takes one meeting to approve the purchase. Spoiler: it never does.

Customs: The Final Boss

You got the quotation approved, the purchase request sent, and the supplier ships with admirable haste. Optimism returns. You tell yourself, “We’ll make the deadline at least. Nothing can go wrong now.” You just had to say it, didn’t you? Cue the email: the equipment is stuck at customs. Why? Because the specs for your shiny new model aren’t on your government’s customs list. Congratulations! You’re the first person in your country to order this hardware. Your prize? Gather every specification, forward it to the Government Regulatory Agency, and let bureaucracy move at its pace. Do I really need to make a witty comment about the speed of it?

Unboxing Day

The day has finally arrived. Your local courier says the delivery is on its way. Like a kid on Christmas morning you unbox the server. Carefully carry it as the entire office watches like you’re holding the crown jewels. You’ve left the perfect spot in the rack—not too high, not too low—just above the UPS and below the routers. It even came with those nifty rails, so you can mount it solo without begging for assistance. You plug one PSU into the UPS, the other into the rack PDU. Then you patch the server to the dedicated ports on core switches and wonder why server has more network ports than a small switch. Hopefully, network team created a VLAN and added routes by now.

Next, you grab the monitor and keyboard from your desk—because who has the budget for a KVM? Your preloaded USB with the OS is in your pocket. You boot it up, ready to install… but first, the server management tool demands configuration. Better obey if you ever want remote access—you can’t leave your monitor on the floor forever.

Finally, the OS installer appears. “Do you agree to the terms of service?” Sure, whatever. “Install locally?” Yes, please. “On which of these six disks?” …Darn it. Power off, reboot, try to interrupt the boot process. Wrong key—it’s not Delete anymore. Try again. This time, you read which key to press. RAID menu. RAID 10. Select the disks. Deep breath. Let’s start the install—again.

The Everest Moment

The OS is finally installing, and you feel like you’ve conquered Everest—except Everest doesn’t reboot three times and ask you to configure a time zone. You set the hostname, which of course must follow the sacred naming convention that nobody documented but everyone swears exists. Then comes networking: static IP, gateway, DNS. You ping the server from your laptop—success! You feel triumphant, like a wizard who just cast a level 8 spell and, against all odds, still kept his original shape. No small feat, considering the last guy who tried it ended up with inexplicable craving for bananas.

Next, you install updates. All 347 of them. The progress bar crawls along slower than paperwork at the Government Regulatory Agency (I couldn’t resist), and somewhere between update 112 and 113, you start questioning your life choices.

Finally, the OS is up, patched, and ready. You install the application—the very reason this whole saga began. Watching the progress bar feels surreal. No RAID menus, no customs paperwork, no surprise reboots—just the sweet, simple act of software sliding into place. And when that final “Installation Complete” message appears, you sit back and think: all this effort, all this chaos… just for a blinking cursor and two words: ‘Hello World.’

Monitoring and Second-Guessing


It’s been several months since the project went live. A few hiccups with app configuration, but otherwise everything’s running smoother than a NOC team coffee machine. As a good engineer, you stay proactive—monitoring logs, watching resource usage. Apart from the occasional spike, CPU sits below 15%, RAM hovers around 40%. You start wondering if you went overboard with the specs. Am I out of touch? No—it’s the developers who are wrong for making an app that can run on a potato.

The Next Big Thing

Then comes the next project: another application, another set of requirements, and another Project Manager with dreams bigger than your rack space. You glance at the server, still humming proudly, and think: “Do I really want to go through customs paperwork again?” No. This time, you decide to play smart—or at least pretend to. If only there were a way to squeeze every drop of resources from your on-prem server.

Squeezing Every Drop of Power

Enter the hypervisor: the intangible layer that promises to turn one physical box into many virtual ones. Your mind starts racing, drafting a task list. First, back up everything and verify the backup actually works—because nothing says “fun” like discovering it doesn’t after you’ve wiped the disks. Then, maybe create a VM from the existing OS. Next comes the point of no return: format the drives, build a new RAID, install the hypervisor, set up virtual networks (and don’t forget to put that network port in trunk mode) and finally restore the VM. Easy peasy, right? Just a week or two of preparation, some testing in staging, and one sleepless night for the actual migration—and you might just pull it off. Otherwise you start updating your resume in the morning.

Did We Just Pull It Off?

You must have done something good in a previous life, because the app is running after migration. Either you fell asleep and are dreaming a blissful dream, or you actually pulled it off. Turns out preparation does pay off. Based on past performance, you tailored resources for the VMs, and now two apps are happily running on separate OS instances—with plenty of capacity to spare.

You triple-check everything before informing management of your heroic endeavor. After all, you saved the company money on a new server, not to mention the power bill. Surely, some kind of reward is earned. Your thoughts wander—visions of bonuses, maybe a plaque—until the CIO replies with a single word: attaboy.

If we were anyone else, we might feel cheated. But we are enlightened, and we know the truth: the skills we honed tonight are reward enough. One more notch on the belt, one more achievement unlocked making us better than we were yesterday. And that, dear reader, is all we can aim for.