IPAM Beyond the Spreadsheet: Managing Address Space Properly
Every network manages its IP addresses. There is no opting out — every device needs one, and something decides which one it gets. The only real question is how that decision happens: deliberately, from records you trust, or by folklore. Folklore sounds like "ask Dave, he remembers which addresses are safe," "the .50s are probably free," "just ping one and see if anything screams." If your honest answer is closer to folklore, you are in good company. Most small and mid-size networks run exactly this way. That lasts right up until the day two devices claim the same address and half the office starts flickering.
This guide is about IPAM — IP address management — done properly, at a scale where you do not need enterprise software or a committee. You will learn what IPAM actually consists of (it is smaller and more concrete than the acronym suggests). You will learn why the trusty spreadsheet always drifts away from reality no matter how disciplined you are. You will learn the one habit — reconciliation — that fixes the drift regardless of what tool you use. At the end you get a starter kit: three small registers you can copy today and a fifteen-minute weekly ritual that keeps them true.
In this guide
- What IPAM actually is: three registers, one agreement
- Why the spreadsheet always drifts
- The reconciliation loop
- The assignment rule
- Reservation policy: carving up the space
- Tooling, honestly
- What good IPAM pays you back
- The IPAM starter kit
What IPAM actually is: three registers, one agreement
Strip away the vendor gloss and IPAM is three linked registers — three lists that describe your address space from three different angles. Think of how a city keeps track of land. There is a zoning map that says what each block is for. There is a land registry that says who owns each individual plot. There is a census that says who actually lives there right now. A city where those three disagree is a city where you get building permits for occupied lots. A network is no different.
- The address plan is your zoning map: which subnets exist, what each one is for, and why. "10.1.20.0/24 is the server subnet at the main site; 10.1.50.0/24 is guest Wi-Fi." One row per subnet. This is the strategic view — it changes rarely, and it answers "where should this new thing live?"
- The assignment register is your land registry: which individual address belongs to what, within each subnet. "10.1.20.11 is the file server; 10.1.20.14 is the backup appliance, owned by us, installed last spring." This is the tactical view, and it is the register people mean when they say "the IP spreadsheet."
- Dynamic reality is the census. It includes what DHCP — the Dynamic Host Configuration Protocol — is actually leasing right now. DHCP is the service that hands out addresses automatically to devices that ask. The census also includes every static address that is actually answering on the wire. You do not write this register by hand; the network writes it. Your job is to read it.
IPAM, the discipline, is simply this: keep all three registers true, and keep them agreeing with each other. Every address-related disaster you will ever troubleshoot is a disagreement between two of these registers that nobody noticed in time. That includes conflicts, exhausted pools, and the mystery device on a server address. Everything else in this guide is technique for noticing.
Why the spreadsheet always drifts
The IP spreadsheet is not a bad idea. Written down, in one place, sorted by address — that is genuinely most of the battle. A well-kept spreadsheet beats an abandoned enterprise tool every single time. The problem is not the format. The problem is that a spreadsheet has no connection to reality. It records what someone intended at the moment they typed it. The network, meanwhile, keeps changing whether anyone types or not. A spreadsheet is a diary, not a sensor.
Worse, the incentives are lopsided. Call it the update-burden asymmetry. Writing to the spreadsheet is a chore that interrupts real work and benefits some future stranger. Reading from it assumes every past stranger did their chore. Updating it is nobody's emergency; trusting it is everybody's habit. So entries get skipped on busy days — which is to say, on exactly the days when the most changes happen. Drift is not a moral failure of your team. It is the natural physics of a document that only changes when a human remembers to change it.
The drift shows up as two classic lies, and each one has its own outage story. The first lie is the "free" address that isn't. The spreadsheet shows 10.1.20.57 as unused, so you give it to the new application server. Except .57 was quietly taken two years ago by the standby firewall's secondary interface — a device that only speaks up during a failover. The new server works fine for months. Then a failover happens, two devices claim one address, and you spend an evening chasing symptoms that make no sense. (The full anatomy of that evening is covered in detecting and resolving IP conflicts — it is a genre of misery all its own.)
The second lie is the mirror image: the static assignment nobody recorded. A vendor tech installs a badge-door controller, types a static address into it because that is what vendor techs do, and leaves. The spreadsheet never hears about it. The address happens to sit inside the DHCP pool, so months later DHCP innocently leases it to a laptop. Now the doors "randomly" stop working every few days. There is a quieter third lie worth knowing too: the ghost entry, an address marked used by a machine that was decommissioned years ago. Ghosts do not cause outages; they cause false scarcity. A subnet looks nearly full and triggers a panicked expansion project when a third of its "used" addresses are actually headstones.
The reconciliation loop
Here is the discipline that fixes all of this. It works whether your register lives in a spreadsheet, a wiki, or a dedicated tool: periodically compare what the register says against what the network says, and chase every disagreement. That is the whole idea. Your register records intentions; discovery collects facts; reconciliation is the meeting where they argue it out.
The facts come from two cheap sources. The first is an active sweep of the subnet: ping every address and see what answers, then read the ARP table. ARP (Address Resolution Protocol) is the mechanism devices use to translate an IP address into a hardware MAC address. They do this before they can talk on the local network. The ARP table is the list of translations a machine has recently performed. It has a lovely property: even a device whose firewall silently drops pings still has to answer ARP to function at all. So sweeping a subnet and then reading the ARP table catches shy devices the ping alone would miss. On your own machine, that looks like this:
# Windows: sweep a /24 you are connected to, then read the ARP table. # The pings populate the table; arp -a shows who is really there. for /L %i in (1,1,254) do @ping -n 1 -w 50 10.1.20.%i > nul arp -a
For subnets you are not plugged into, read the ARP table of the router that serves them instead — over SNMP or its admin interface. The router has been resolving addresses on that subnet all day and its table is the honest census. The second fact source is the DHCP lease table. It is the server's own list of which addresses it has handed to which devices, complete with hostnames and MAC addresses. Between a sweep and the lease table you have dynamic reality in writing. (Discovery is a deep and useful subject on its own. The Network Discovery & Asset Inventory guides cover the methods properly. If you have inherited a network where nothing is documented at all, the five-day discovery sprint is the crash-course version.)
Now diff the facts against the register. Every address lands in one of three buckets, and each bucket has a standard move:
| Finding | What it means | What to do |
|---|---|---|
| In register, seen on network | All is well | Nothing. Enjoy the moment. |
| In register, not seen | A ghost: decommissioned, moved, or just powered off today | Mark "unseen" with the date. Reclaim only after it stays unseen past a quarantine period (30 days is sane) — laptops travel and servers reboot. |
| Seen, not in register | A squatter: an unrecorded static, or a lease outside your notes | Identify it (MAC vendor, hostname, switch port), find the owner, and either record it properly or evict it. |
The first reconciliation on a neglected network is always humbling — expect a parade of ghosts and squatters, and budget real time for the chase. The second one is smaller. By the fourth or fifth, reconciliation takes fifteen minutes and finds one or two drips, which is the whole point. You have converted an invisible, compounding error into a small, boring chore. Small and boring is what winning looks like in address management.
Tip: even with a trusted register, verify before you assign. Ping the address, check ARP, then configure it. It costs ten seconds and it is the difference between finding the squatter now, politely, and finding it later, during the outage it causes. Trust, but ping.
The assignment rule
Reconciliation heals drift after it happens. The assignment rule slows the drift at its source, and it is one sentence: no static address exists until its register entry exists. Not "record it when things calm down." Not "I'll do a batch on Friday." The entry is part of the assignment, the way the label is part of the cable. If it is not written down, it is not assigned — it is squatting with your permission.
The way to make this stick is to invert the flow. Stop treating the register as a place you report addresses to. Start treating it as the place you get addresses from. When you need a static, you open the register and look at the correct range for that kind of device (next section). You take the next free address and fill in the row — address, hostname, purpose, owner, date. You do all of that before you touch the device. Sixty seconds. Done this way, the register cannot lag reality, because the register moves first. This is the intake habit, and it is the single highest-value process change in this entire guide. It also needs to bind everyone who assigns addresses. That includes the colleague who "just borrowed one real quick" and, especially, any vendor tech who visits with a laptop and confidence.
Reservation policy: carving up the space
The assignment rule gets easier when each subnet has a published internal layout — ranges carved by purpose. Then choosing an address is a lookup instead of archaeology. Without a carve, every assignment is a little research project ("is anything near .80 important?"). With one, the question answers itself: infrastructure goes here, servers go there, DHCP owns that stretch, and nothing else may live inside it. A typical carve for a /24 looks like this:
| Range | Purpose | Notes |
|---|---|---|
| .1 | Gateway | Always. Every subnet, no exceptions — findability under stress. |
| .2 – .19 | Network infrastructure | Switches, access points, UPS units. Static. |
| .20 – .49 | Servers and static services | Static or DHCP reservations, every one in the register. |
| .50 – .199 | DHCP pool | The server leases these. No statics inside, ever. |
| .200 – .239 | Reserved for growth | Deliberately empty. Future pool expansion or new statics. |
| .240 – .254 | Test and temporary | Lab gear, staging. Reviewed at every reconcile. |
Two details in that carve prevent whole categories of outage. First, statics never live inside the DHCP pool range. The pool belongs to the DHCP server alone, which makes the unrecorded-static-versus-innocent-lease collision structurally impossible instead of merely discouraged. Second, the growth reserve is real, labeled, and empty. When the pool runs hot you widen it into space you set aside on purpose, calmly, instead of negotiating with whatever happens to be adjacent. The exact ranges matter far less than their existence and their consistency. Use the same carve in every subnet and your future self can navigate any of them from memory.
Tooling, honestly
Now the question everyone starts with, deliberately answered last: what tool should you use? Honestly — it depends on scale, and there is a ladder. What does not depend on scale is the discipline above; a tool automates reconciliation, but no tool absolves you of it.
Rung one: a disciplined spreadsheet plus the reconciliation loop. For a single site with a handful of subnets, this is a perfectly workable floor, and this guide's closing templates give you the exact sheets. The spreadsheet's fatal flaw was never the format — it was running blind, with no reconciliation against reality. Add the loop and the flaw is gone; skip the loop and no software rung above will save you either.
Rung two: the IPAM features inside tools you already run. Your DHCP server already maintains lease truth and scope statistics. Your monitoring platform likely offers subnet views, discovery scans, and utilization tracking. Many DNS servers track address usage too. Before buying anything, inventory what you already own. Most small networks discover they have two-thirds of an IPAM system installed and nobody had introduced the pieces to each other. This middle rung, wired together, comfortably carries most networks up to a few dozen subnets.
Rung three: a dedicated IPAM suite. You may reach multiple sites, overlapping teams, or delegation (letting the branch admin manage the branch's ranges without touching yours). Or you may reach automation that needs to request addresses through an API. In those cases, the middle rung starts to strain and purpose-built IPAM earns its keep. That is a real threshold, and smaller networks genuinely do not need to cross it early.
Whatever rung you shop on, judge candidates by criteria, not brand names — brands churn, criteria do not. The ones that matter: it reconciles against reality (built-in discovery, or clean imports of sweep and lease data). It shows utilization per range at a glance, so "are we running out?" is a look-up, not a project. It records who, when, and why for each assignment. And it exports to something plain like CSV, because your data must outlive the tool. A tool missing the first criterion is a prettier spreadsheet — it will drift with exactly the same physics, just with nicer fonts.
What good IPAM pays you back
All of this is overhead until the day it isn't, so it is worth naming what the discipline actually buys. First, and most often: troubleshooting speed. Half of network diagnosis begins with "what IS 10.1.20.31?" That question might start with a firewall log entry, a strange ARP neighbor, or an alert about an address you do not recognize. With a true register that question takes ten seconds; without one it takes an hour of tracing, during an incident, with someone standing behind you. Second: collision-free provisioning — new machines get addresses that are genuinely free, and the conflict genre of outage quietly disappears from your life. Third: audit answers. A security review or an auditor asks "what is on this network segment and who owns each address?" You export a sheet instead of scheduling a discovery project. And fourth, the biggest and least visible payoff: the renumbering you never do. Address plans laid out with headroom and clean boundaries grow gracefully for years. Plans made ad hoc eventually corner you into readdressing live systems, which is the project everyone underestimates. That story — and the planning habits that let you avoid starring in it — is the subject of subnet planning for growing networks.
The IPAM starter kit
Here is everything above compressed into something you can build this week. Three registers, then one ritual. The address plan — one row per subnet, kept wherever your team already keeps documents:
| Network | Purpose | VLAN | Gateway | DHCP range | Notes |
|---|---|---|---|---|---|
| 10.1.10.0/24 | Users — main site | 10 | 10.1.10.1 | .50 – .199 | Growth reserve .200–.239 |
| 10.1.20.0/24 | Servers | 20 | 10.1.20.1 | none | Statics .20–.49 only |
| 10.1.50.0/24 | Guest Wi-Fi | 50 | 10.1.50.1 | .50 – .199 | Short leases; isolated |
The assignment register — one sheet or section per subnet, one row per static or reservation. The "why" column looks optional; it is the column your successor will bless you for:
| Address | Hostname | Purpose / why | Owner | Assigned | Last seen |
|---|---|---|---|---|---|
| 10.1.20.11 | fs01 | File server | IT | (date) | (reconcile fills this) |
| 10.1.20.14 | bkp01 | Backup appliance — vendor-managed | IT / vendor | (date) | (reconcile fills this) |
The third register — dynamic reality — you do not maintain by hand: the DHCP lease table and your sweep results are that register. What you keep is a thin reconciliation log so drift becomes visible over time:
| Date | Subnet checked | Ghosts | Squatters | Resolved? |
|---|---|---|---|---|
| (date) | 10.1.20.0/24 | 1 (.33 — old print server) | 1 (.41 — unrecorded NAS) | .33 quarantined, .41 registered |
And the weekly reconcile ritual, fifteen minutes, same coffee slot every week:
- 1. Open the DHCP console: any scope above 80% used? Any server warnings? (Two minutes — and the early-warning half of this is worth automating; see monitoring DHCP scopes.)
- 2. Sweep one or two subnets — rotate through them week by week — and pull the ARP results.
- 3. Diff against the assignment register: list ghosts and squatters.
- 4. Chase squatters to an owner: register them properly or evict them.
- 5. Date-stamp ghosts; reclaim any that have stayed unseen past your quarantine window.
- 6. Log one line in the reconcile log. Close the sheet. You are done.
Tip: put the ritual on the calendar as a recurring appointment with yourself, not a to-do item. To-do items lose fights with tickets; appointments mostly don't. A reconciliation habit that runs 45 weeks a year beats a perfect one that ran twice.
Where to go next
You now have the honest shape of IPAM: three registers — plan, assignments, dynamic reality — and the reconciliation loop that keeps them agreeing. Start with the templates, adopt the assignment rule today, and let the weekly ritual grind the drift out of your records. From here, the natural next steps are designing the address plan itself with room to grow, and watching your DHCP pools before they fill. Another step is turning the whole thing into documentation that survives staff changes. The rest of the IP Address Management & Subnet Monitoring guides take each of those in turn. And documenting address space is where the three registers grow into a full handover-proof document set.
Put this into practice with SysPeep Monitor
SysPeep Monitor is not an IPAM suite, but it hands you the half of IPAM most tools skip: reality. Its auto-discovery finds the devices actually living on each subnet. So your register has live facts to reconcile against instead of memories. The free trial installs on any Windows machine in minutes.
Download Free TrialLearn About SysPeep