01
Define what each quote actually includes
For dedicated rental, record the chassis specification, term, network inclusion, replacement responsibility, support boundary and cancellation rule. For colocation, record cabinet or rack units, usable power and circuit design, cooling assumptions, facility access, carrier or transit service, IP resources, cross-connects and support response. A space-and-power quote alone is not a complete operating server.
Use written quotes for the intended site and contract date. Colocation products differ by facility, cabinet and provider; the presence of a service at one data centre does not make it universal. Likewise, do not assume a rental operator manages the operating system merely because it replaces failed hardware.
02
Inventory the full colocation lifecycle
Start with hardware purchase, memory and drives, rails, cables, spare parts, shipping, insurance, tax and staging labour. Add recurring space, measured or committed power, connectivity, addresses, cross-connects and monitoring. Then add remote-hands tasks, travel, failed-part shipping, warranty handling, secure disposal and removal at contract end.
Cross-connects are dedicated physical links between parties inside a facility and may be ordered separately from the cabinet. Remote-hands services cover physical activities such as installation, component replacement, cabling and power cycling when remote access is insufficient. Treat both as scoped services with their own quote and turnaround conditions, not as invisible facility features.
03
Assign responsibility before an incident
Build a responsibility matrix for detection, authorization, physical access, replacement stock, data-bearing media and return to service. With rental, the provider normally owns the physical asset while the customer remains responsible for software and data unless a managed service says otherwise. With colocation, the customer owns the equipment and must arrange the people and parts needed to restore it.
The matrix should name escalation contacts and approval limits. An emergency that needs a drive, compatible cable or console session is a poor time to discover that no spare is on site or that a technician lacks precise instructions.
Assign responsibility before an incident| Activity | Dedicated rental question | Colocation question |
|---|
| Failed component | What does the rental SLA or policy cover? | Who supplies, ships and installs the spare? |
|---|
| Console access | What out-of-band path is included? | Is customer OOB equipment installed and secured? |
|---|
| Network handoff | Which port and addresses are included? | Which transit, carrier and cross-connect are contracted? |
|---|
| Exit | When can service end? | Who removes, ships or disposes of assets? |
|---|
04
Model capital, financing and exit explicitly
Set an exact analysis horizon and discount or financing method. For owned equipment, state purchase date, expected service life, warranty, residual value and refresh assumption. Residual value is not cash until the asset is sold, so include inspection, de-racking, freight and disposal work. For rental, include every renewal over the same elapsed days and any migration overlap.
Run at least an expected case and stress cases for higher power, an early component failure, extra remote work and early exit. Do not invent failure rates or resale values; use quotes or leave the variable visible. Colocation can be attractive when hardware customization and long stable use justify ownership, while rental can be attractive when low initial capital and a simpler hardware boundary matter.
05
Compare outcomes, not just rack price
The expected result is a sourced cost ledger, responsibility matrix, incident path and exit plan for each option. Keep qualitative differences beside the total: component freedom, provisioning lead time, physical-access burden and ability to change capacity. The lower number is credible only when both sides deliver the same recovery and connectivity requirements.
Direct answers
Questions about this guide
Is colocation cheaper once the hardware is paid for?
Not necessarily. Space, power, connectivity, cross-connects, remote work, spares, financing and exit continue. Compare a complete quoted lifecycle over the same horizon rather than treating the chassis as the only colocation cost.
Does dedicated rental include operating-system management?
Do not assume it. Hardware ownership and replacement are different from patching, application operations, backups and recovery. Use the provider's written service scope.