Colocation or managed private cloud for hardware you already own
Kestrelmoor Surveying keeps its project files, its accounting system and the license server for its survey software on four servers in a closet at its main office. Three of them were bought in March 2024 with five years of manufacturer support, so their contracts run to March 2029. The fourth dates from 2019, runs the accounting system, and its support contract has already lapsed. Next June the office lease ends, the new space has no server room, and the two people who make up the IT team want the closet's cooling alarm gone. Renting capacity from a cloud would leave three servers with about 21 months of support unused. Shipping all four to a data center keeps them working, but someone still has to run them once they get there. Three setups answer that differently, and two things decide between them: how much life each server has left, and how much of the operating work your staff want to keep.
How much life the hardware has left
Make one row for each server and storage device. Record the model, the purchase date, the date its support contract ends and what runs on it.
Finance can tell you how the servers are carried. IRS Publication 946 treats a computer as five year property, so a company that depreciates its servers over that period may still be depreciating 2024 hardware in 2028, unless it took Section 179 or bonus depreciation. The schedule says nothing about whether a server still works, and finance can tell you what retiring one early would cost on paper.
Then sort the rows. Hardware with years of support left is worth moving, because the support you paid for stays usable in the new rack. A server near or past its support end is a poor candidate for the trip, because replacing it means a second move with the same planning. Kestrelmoor's three 2024 servers belong in the first group, and its 2019 server belongs in the second.
Three setups for the same servers
At its simplest, colocation means renting rack space, power, cooling and a network connection in a data center. Your servers move in, and your staff keep doing everything they did in the closet, from a distance. In the second setup the same servers move in, and the provider's engineers maintain and monitor them under a contract that lists the tasks. The third moves the workloads onto virtual machines on the provider's hardware, while your servers are retired, sold or kept as spares.
| Task | Colocation your staff run | Colocation the provider maintains | Workloads on the provider's hardware |
|---|---|---|---|
| Racking and cabling on moving day | Your staff, or remote hands you order | The provider, where the contract says so | The provider, which also moves the workloads |
| Firmware and host patches | Your staff | The provider, for the tasks the contract names | The provider |
| Hardware alerts at night | Your staff, from a monitoring tool you run | The provider, if the contract lists it | The provider |
| A failed disk or power supply | Remote hands swap the part and your staff do the rest | The provider, within the contract's response time | The provider |
| Buying the next servers | You | You | The provider |
| Guest operating systems and applications | Your staff | Your staff, unless the contract adds them | Your staff, unless the contract adds them |
The middle column depends on the contract. "Maintained and monitored" can mean a provider replaces parts and patches firmware, or only that someone watches a dashboard and calls you. Ask for the list by task.
When a disk fails after the move
A failed drive in the closet meant a walk down the hall. In a data center it means a part shipped to a new address, and a person with permission to receive it and swap it.
Before the servers move, update the service address on each manufacturer support contract so a replacement part goes to the data center instead of the empty office. A spare drive and a spare power supply of each type you run, kept in the rack, let a swap happen before the shipment arrives. Give the data center a written list of who may sign for deliveries addressed to your company.
Decide what happens to the failed drive, too. Dell sells Keep Your Hard Drive for Enterprise, which lets a customer keep a failed drive and control how it is disposed of, and the service has to be bought before a drive fails. Ask your manufacturer whether it offers anything similar, because a drive full of project files should not leave your custody in a return box.
When your own staff run the servers, the guide to cloud repatriation for teams under 50 engineers shows where remote hands stop and your engineers take over. Under a maintenance contract, ask the provider to name the person or team for each step, from ordering the part to watching the array rebuild, with a response time for each. On the provider's own hardware, the provider handles the failed drive, so ask for its disposal terms in writing.
Licenses tied to a server
Some licenses follow the hardware instead of the workload. Dell's article on its OEM Windows Server licenses says they are sold bundled with the hardware and can be used only on that specific hardware, and Dell attaches the license to the server chassis. When your staff run colocated servers, the license moves with the chassis. If the workloads move onto a provider's hardware, those licenses stay behind with servers that no longer run anything.
Licenses bought through Microsoft Volume Licensing can move, within limits. Microsoft's Universal License Terms let a customer reassign a license to another device, though not within 90 days of that license's last reassignment unless the reason is permanent hardware failure or loss, and the software has to come off the former device. The same terms bar using the software on devices under a third party's management or control, except where Software Assurance or a subscription grants outsourcing rights. That limit can matter for the second and third setups, so your reseller can confirm what your licenses allow, and the provider can say whether it supplies licenses for its own hardware. The answers change what those setups cost.
Line of business software needs the same check. Kestrelmoor's survey software reads its license from a USB key plugged into one server, so the move plan has to say where that key lives afterward and whether the software still finds it on a virtual machine. If a hypervisor renewal is open at the same time, the guide to a VMware renewal quote for three hosts recalculates the quote from the cores your hosts run.
Planning the next refresh
Colocation postpones the hardware question until the support contracts end. When Kestrelmoor's three servers reach March 2029, someone has to buy their replacements, ship and rack them, move the workloads across and retire the old machines, all in a rack the IT team rarely sees.
Agree on that refresh before signing. Find out from the provider whether colocated workloads can later move onto its own hardware and what that move involves, because the third setup may suit servers that have aged out.
Retired drives need an owner too. NIST's Guidelines for Media Sanitization (SP 800-88 Revision 2, September 2025) describe sanitization as a process that makes access to the data on the media infeasible for a given level of effort, and they help organizations set up a sanitization and disposal program matched to how sensitive their information is. Decide who sanitizes drives that leave the rack, by what method, and what record comes back for each serial number.
Kestrelmoor split the decision. Its three 2024 servers go into colocation under a maintenance contract until March 2029. The accounting system moves onto a virtual machine on the provider's hardware, so the 2019 server never makes the trip.
List the servers, then compare
Put the hardware list beside the operating work. The operating work sheet records who does each of 25 recurring tasks, such as patching and restore tests, and the hours each one takes. Fill it in for today, then mark which tasks stay with your staff under each setup. Start with the tasks nobody does today.
Send each provider the same questions in writing. Ask what maintenance and monitoring cover, task by task, how colocated hardware is kept apart from other customers' networks, whether workloads can move onto its hardware later and who sanitizes drives when hardware retires. The due diligence questions for a managed private cloud provider cover the rest of the contract.
We own the hardware Petrichor runs on, and we also offer colocation for your own segregated hardware, managed by your team or with Trinsic, our parent company, supporting it. Talk to a consultant with your server list, and pick Virtual servers under "What do you need help with?" on the form.