The Internet's Spare Parts Economy: Why Your Network's Most Valuable Asset Might Be the Equipment Sitting on Your Shelf
The expensive thing I ever lost was a fifty-dollar part I had never purchased.
I did not lose it in a fire or a flood. I lost it in the way many network people lose items: by deciding, calmly and for good reasons, that I did not need it yet.
The part was a power supply for a switch that had worked perfectly for six years. Nobody thought about it. Then, on a Friday afternoon, it stopped working, and service for a third of an ISP's customers went down. The switch itself was fine. The network design was fine. One tiny component, which a supplier could have sold me for the price of a dinner, turned a quiet Friday into most of a weekend of network downtime.
That weekend changed the way I see the shelf in the room. For years I looked at it. Saw clutter, boxes, and money I could not use. Now I see it as one of the valuable things in the building.
This post is about that shelf. It is about the economics of spare parts, why spare parts are easy to underrate, and how an ISP of any size can build a sensible hardware inventory strategy without drowning in boxes. I will tell you what went wrong for me, what I realized after it went wrong, and what I would do differently if I started again tomorrow.
The Part That Cost Me a Weekend
Let me tell you how that Friday actually went, because the details matter more than the headline.
It was 4:40 in the afternoon when the alerts started. One aggregation switch, the device that gathers traffic from a neighborhood of customers, had gone dark. My first thought was a fiber cut. My second thought was a software crash. It took twenty minutes of poking before someone simply walked over and looked at it. The power supply light was off. That was all. No drama, a dead box that cost less than a good pair of work boots.
Here is where it got ugly. The switch had a slot for a power supply and a backup, and that slot was empty. Years earlier, someone had said, "We will add the one later." Later never came.
I went looking for a replacement. The local distributor was out of stock. The manufacturer had stopped selling that model, and the newer version used a different connector. A colleague at another company had one. It was a three-hour drive, and he was away until Monday. In the end, we paid for rush shipping from another country, then waited, then fought a configuration problem when it arrived.
Somewhere around Saturday night, I did the math in my head. Felt sick. We were crediting customers paying overtime, answering calls, and watching a few people sign up with a competitor. The total came to a number I would rather not repeat. It was times the price of the shelf I had refused to stock.
The failure itself was not the problem. Things break. The problem was that I had no parts and no options on a Friday afternoon. A failure with a spare on the shelf is a twenty-minute chore. A failure without one is a crisis, and the only difference between them is a decision made months earlier.
Why Spare Parts Feel Like Waste
Before I judge myself too harshly, I should admit that his reasoning was not stupid. It followed the logic that most businesses use.
Think about how it looks on paper. It is money that I have already spent on something that is not earning anything. It sits in a box. It slowly gets older. It takes up space that could hold something. Anyone who runs a budget will look at that and ask a fair question: why pay for equipment that might never be used?
Here is the trap. A spare part has a cost that I can see and a benefit that I cannot. I can see the invoice. I can see the box. I cannot see the outage that did not happen because it did not happen. Nobody writes a report titled "The Weekend We Stayed Online Because of a Fifty-Dollar Purchase."
The opposite also holds. Network downtime rarely appears as a clean line in the accounts. It shows up as credits, overtime, lost sign‑ups, a team, and a few customers who quietly leave. Those costs are scattered, so they are easy to blame on luck instead of a decision.
There is one belief hidden underneath all this, and it is the most dangerous one: "If something breaks, we will just order another." That sentence assumes the part is in stock somewhere, the price has not jumped, the shipping is fast, and the new unit works with what you have. Every one of those assumptions failed for me in a weekend.
In my experience, ordering a part only after it breaks is not a strategy. It is a bet that the rest of the world will cooperate on your day. Sometimes the bet pays off. When the bet does not pay off, you pay for it times over.
The Math Nobody Does
After that weekend, I did something I should have done years before. I sat down. Wrote out what an outage actually costs, using my own numbers instead of gut feeling.
You can do the same in ten minutes. For any piece of equipment, ask:
How many customers go offline if this one device fails?
What do you pay out in credits, overtime, courier fees, and truck rolls (trips to fix things in the field)?
How long would you realistically wait for a replacement if you had to order it today?
What happens to the customers who leave and never come back?
Now compare that total with the price of a spare. For pieces of equipment, the comparison is almost embarrassing. A spare that costs a hundred dollars only needs to stop one bad weekend once over the whole life of the device to pay for itself.
I want to be honest about the side too because it matters. You should not keep a spare for everything. That is how people end up with a back room of expensive boxes and a cash problem. The goal is not to own equipment. The goal is to own the equipment. I use a three‑part filter:
Likelihood. Is this a part that tends to fail, like a power supply, a fan, or a plug-in optical module?
Impact. If it fails, how many customers lose service? One home is annoying. A whole neighborhood is serious.
Wait time. How long is the lead time to replace it? A part you can buy across town tomorrow needs a spare far less than one that ships from overseas in six weeks.
Things that score high on all three get a spare. Things that score low on all three do not. Most of the decisions lie in the middle, and that is where your own experience and your own failure records should guide you.
This is also why spares are an investment rather than a simple expense. An expense disappears. A spare sits there quietly, protecting your network reliability, your reputation, and your team's weekends. It keeps its value as long as it is tested and ready.
Standardize Before You Stockpile
Here is a mistake I made after that weekend that cost me a kind of money. In my panic to never be caught out again. I started buying spares for everything. Then one afternoon I counted what we were actually running.
We had twenty‑three models of customer routers in the field. Five different types of plug‑in modules that connect fibre to a switch. Three brands of power supplies that did not fit each other. Every one of those required its own settings and its own page, in someone’s memory.
I had built a network that was impossible to protect. You cannot keep a spare for twenty‑three items without burying yourself.
That is when equipment standardization became my boring idea. The principle is simple: run kinds of equipment. When you have four router models of twenty‑three, one spare can cover hundreds of customers. When most of your switches belong to the family, a replacement can go into any of them with the same settings file. A technician who has had a night does not have to remember which box is which.
The benefits stack up quickly:
A shelf because each spare covers more of your network.
Faster swaps, because the steps are the same every time.
Easier training, because a new hire learns four things of twenty-three.
Simpler ordering, which also gives you weight when negotiating with suppliers.
Now, a warning, because I have seen people overcorrect. Standardizing on one brand forever can lead to a vendor lock-in trap, where one maker's delays or price rises become ISP's problems. The smarter approach is to standardize on standards: choose a set of equipment that follows open, shared rules and keep a second approved brand in your back pocket that ISP has already tested on its network.
This is a multi‑vendor approach with discipline. ISP keeps the simplicity of a list, and ISP keeps the freedom to move when the market forces ISPs hand. I would rather have two tested options than one option.
Does Every ISP Need a Hardware Inventory Strategy?
Short answer: yes. Strategy does not have to mean a warehouse, a software system or a full‑time stock manager. For an ISP, it can fit on one page.
What changes with size is the scale of the plan, not the need for one. Here is how I would think about it.
If ISP is an ISP with a few hundred or a couple of thousand customers ISP plan is a short list. ISP picks the handful of devices where one failure would hurt most, and ISP keeps one spare of each. ISP adds the boring parts that fail often: power supplies, patch cables, fiber modules, and a few customer routers. ISP then writes down who has the key to the cupboard. That alone puts ISP ahead of most of the networks I have seen.
If ISP's a regional ISP, it probably has more locations, and travel time starts to matter. A spare in the head office does not help a tower two hours away. At this stage, ISP adds a stock point, a simple tracking sheet and a rule for restocking whenever someone uses a spare. The rule matters more than the spreadsheet. A spare that gets used and never replaced is a delay in the next crisis.
If ISP is an operator, ISP is dealing with several sites, a mix of old and new equipment and real money tied up in stock. Here the question becomes how to hold less without being exposed. That is where agreement occasionally is needed.
Across all three sizes, I would push ISP toward one idea: with suppliers coming in, such as set-aside stock, fast replacement terms, or help holding parts. ISP only: protect every point of failure. That is the one device whose failure takes a group of customers offline. Find the ISP's point. Most networks have more of them than the owner thinks, and they are usually the devices nobody looks at because they have worked well for so long.
I will add one caution from experience. Spares can go bad. Hardware gets discontinued; firmware falls behind. A box that sat for five years may not match what ISP runs today. Overstocking can hurt as much as understocking, only more slowly. The aim is a shelf that's small, current, and useful.
Predicting What Will Break Next
Once the ISP's shelf was small and sensible, the ISP wanted to stop guessing. That is where ISP started learning what people call predictive spare‑part planning. The name sounds fancy. The idea is not. It means using ISPs history to decide what to keep instead of using fear.
ISP does not need software. A basic spreadsheet is enough. For each part, ISP tracks four things:
How many are in service.
How many have failed in the year or two.
How long it takes to get a replacement.
How many spares ISP currently holds.
Then ISP applies one rule. The number of spares ISP needs is roughly the number of failures ISP expects while ISP is waiting for a replacement, plus a small safety margin.
Here is a worked example with numbers to explain the method. Suppose a power supply fails about twice a year in your network, and it takes twelve weeks to replace it. Twelve weeks is below a quarter of a year, so you would expect about half a failure while waiting. One spare fits that expectation comfortably. If the lead time were twenty‑four weeks, you would expect one failure during the wait, and a second spare looks wise. The same part, wait, different answer.
A few patterns helped me predict trouble earlier:
Age. Devices bought together fail around the same time, so a group of old units is a warning even if all run fine now.
Environment. Heat, damp, dust, and power surges shorten life. Equipment in a cabinet or on a stormy hillside fails more often than the same equipment in a cool room.
Season. In my area, summer heat and storm season brought failures. If your pattern looks similar, stock up before the months, not during them.
Notices. When a manufacturer says a product is ending sale, that is your signal to buy parts now while they are still available.
None of this makes you a fortune teller. What it does is turn a guess into a reasonable plan. It gives you something concrete to say when someone asks why you are holding spare stock. Good records also quietly prove that your network reliability did not happen by accident.
Running the Shelf Like a Real Asset
Here is a story I'm not proud of. A couple of years after that weekend a switch failed again. This time I walked to the shelf with a grin, grabbed a spare and plugged it in. Nothing happened. The spare was dead too. It had been sitting in a corner for years, and nobody had ever powered it on to check.
A spare you have never tested is not a spare. It is a hope in a cardboard box.
That day taught me that the shelf must be managed seriously, as the network itself. These habits made the biggest difference for me:
Test every spare when it arrives. Power it on, update its software, and make sure it works before putting it on the shelf.
Pre‑load the settings. If you can keep a configuration for each device type so a swap takes minutes, not hours.
Label everything clearly: what it is, what it replaces, where it was last used, and the date it was tested.
Keep it dry, cool, and clean. A basement corner with a pipe is not good storage.
Count it on a schedule. A quarter check what is on the shelf against what your list says. You will find surprises.
Replace what you use away. The moment a spare leaves the shelf, order its replacement.
Decide who can take things. If anyone can grab a spare without writing it down, it will quietly vanish.
Write the swap steps down. At 2 nobody remembers anything. A one‑page checklist beats an engineer who is half asleep.
The real test is simple. If your best engineer is on holiday and a core device dies tonight, can someone else find a spare trust it and install it? If the answer is yes, you have an asset. If the answer is "maybe", you have a collection of boxes.
Where a Good Supplier Fits In
I do not want to leave you with the idea that the answer is to buy everything and hide it in a cupboard. No ISP, careful, can hold every part it might ever need. Your shelf covers the emergency. Your supplier covers everything after that.
That is why I judge suppliers by a standard, than before. Price still matters. These three things matter more:
Real stock. Not the kind that says "available soon" or "ships in eight weeks". Stock that is in a warehouse and can arrive in a few days. A part that is cheap but only arrives in season is not truly cheap.
Choice across brands. A supplier who gives a real multi‑vendor selection lets you switch when one maker runs out raises the price or stops making a model. The supplier is the way to keep options open.
People who know the equipment. The best question to ask is, "Will this work with what I use?" A supplier who has technical knowledge can answer before you buy, not after you open the box and face the problem.
That mix of stock on hand multiple brands and real technical help is what companies like ISP Supplies are built around. It is the combination I would look for no matter which company I buy from. When your shelf and your supplier work together, the shelf can stay small because you know the next part is a day away and not months.
List the three devices whose failure would hurt your customers the most.
For each device, find the part. Is there a part? Has the spare part been tested? Does the spare part match the part that is installed?
For each part, find out how long a replacement would take if you ordered the part today.
Wherever the answers make you worry, fix that first.
That is all. No budget meeting, no consultant, no software. I think about that part more than I would like to admit. The part never made it onto any list of my valuable assets, and it should have been at the top. Your own version of that part is there on your network right now, working quietly and waiting for its worst day. The only question is whether, when that day arrives, you will reach for a part or, for your phone.
Comments