Preparing for Colocation: A Checklist

Colocation trades your closet's limits for a facility's strengths — conditioned power, real cooling, generator-backed uptime, fast transit. What it does not include is you, standing next to the machine. Every preparation on this list exists to make physical presence unnecessary, because in colo the expensive resource is the drive to the datacenter.

Remote management: the non-negotiable

iDRAC Enterprise or iLO Advanced, configured before shipping: dedicated management port, static addressing that matches the colo's assigned management VLAN, strong unique credentials in the password manager, firmware current. Verify from outside your own network that the console loads and virtual media mounts — this interface is now the only screwdriver you own. If the facility offers an OOB/VPN path to management networks, test that too; if not, a tiny VPN appliance or jump host in the rack earns its 1U forever.

Boot resilience: assume nobody is watching

BIOS set to restore last power state (an outage-recovery must), boot order locked to the mirrored boot device, and that boot genuinely mirrored — BOSS pair or RAID1 — because a dead single boot SSD at 3 a.m. is otherwise a remote-hands ticket with your weekend attached. Disable the boot-interrupting temptations: PXE-first ordering, one-time menus that a power blip can strand. Test the full loop before shipping: power pull, unattended return to service, workloads self-starting.

The physical layer, done once and properly

Correct sliding rails for the cabinet's post type; both cord ends labeled plus asset labels front and rear (remote hands find machines by label, not by your memory); cable lengths measured to the actual cabinet's PDU and switch positions; blanking panels for your unused U-space, which shared-cabinet neighbors and airflow both appreciate. Photograph everything cabled before the doors close — the picture answers half of all future remote-hands questions.

Power and spares arithmetic

Know your committed amps and your measured draw (metered PDU data from burn-in); dual PSUs land on the A and B feeds, and staying under 80% of commitment on one feed means a feed failure does not become a breaker trip. Spares strategy per distance: a cold-spare PSU and boot drives taped inside your own cabinet space cost nothing and turn common failures into 15-minute remote-hands swaps; a whole cold-spare chassis is a rational line item once the drive time exceeds an hour.

Day one at the facility

Rack, cable, then prove every remote path before leaving the building: console, virtual media, power cycling via BMC, monitoring green, alerting to your phone, one full reboot performed from the parking lot. The deployments that go smoothly are the ones where leaving the facility changes nothing you can do to the machine — that is the whole checklist, compressed.