Data Center Consolidation

What this covers

Many agencies still operate server rooms in office buildings, courthouses, or public works yards. Consolidating them, whether into a central agency data center, a colocation facility, or the cloud, reduces risk and cost. The hard part is doing it without disrupting the departments that depend on those systems.

Typical activities

• Physical and logical inventory of servers, storage, network gear, and the applications on them
• Dependency mapping, including the systems nobody remembers until they go down
• Target-state design and a business case comparing colocation, central hosting, and cloud
• Migration sequencing that respects fiscal year timing, election cycles, and seasonal peaks
• Decommissioning with data sanitization records and asset disposition

What you receive

An inventory and dependency map, a consolidation plan with cost comparison, migration runbooks, and decommissioning records suitable for audit.

Considerations for public agencies

Surplus property rules, records retention, and grant-funded equipment restrictions all affect how equipment can be retired. We build those requirements into the plan rather than discovering them at the end.

Signals that consolidation is worth scoping now

• A colocation agreement, lease, or circuit is approaching term, and the notice period to avoid automatic renewal falls well before it.
• Hardware has passed vendor end of support, or spare parts now come from the secondary market.
• No headroom is left: no usable rack space, a portable cooling unit in service, UPS batteries past their rated life, a generator never sized for the room.
• The room is in the wrong building: leased space, a floor scheduled for renovation, a basement with flood exposure.
• Hypervisor or storage licensing terms have changed and a renewal decision is due anyway.
• An audit or insurance review has flagged physical access, environmental monitoring, or the absence of a tested restore.
• One person understands how the environment fits together, and that person is close to retirement.

Any one can be handled in place. Several in the same budget cycle changes the question from repairing the room to whether it should exist.

What the agency supplies

• Read-only access to hypervisor, backup, monitoring, and DNS systems, and badge access to every room holding equipment. Discovery otherwise stops at what tooling reaches.
• Fixed asset records and the purchasing file for maintenance agreements and circuits, whose notice and termination dates constrain the schedule more than the technical work does.
• A named application owner in each department. Tooling shows traffic; owners explain scheduled jobs, file shares, and interfaces that run once a quarter.
• Loaded cost inputs from finance: facility and power charges, depreciation schedules, and staff time, so the status quo is priced like the alternatives.
• Approved downtime windows, plus validation testers, help desk staging, and one person with authority to call a rollback while a window is open.
• Sign-off from whoever holds property, records, and security authority, obtained before equipment leaves the building rather than after.

Downtime windows are the input most likely to hold the schedule, since they depend on departmental availability rather than engineering effort.

Requirements worth writing into the solicitation

• The inventory and dependency map delivered in a machine-readable format the agency keeps, not entries in a vendor portal.
• The discovery method named: agent-based collection, flow analysis, interviews, or a combination, with the coverage gaps each one leaves.
• A business case accounting for exit costs, data egress, and remaining contract term on each option.
• Per-wave runbooks with rollback criteria, validation steps written by application owners, and a rehearsed move group before production systems move.
• Media sanitization to a recognized guideline, with certificates identifying each device by asset tag and serial number. NIST Special Publication 800-88 is commonly cited and describes clear, purge, and destroy as sanitization categories.
• Handling terms for regulated workloads, such as criminal justice information under the FBI CJIS Security Policy and federal tax information under IRS Publication 1075, naming the version in force since both are revised periodically.
• Contract and circuit termination dates tracked as a deliverable, so notice windows are not missed while attention is on the moves.
• Assessment and planning priced separately from execution, since until the inventory exists, execution scope is an estimate priced against systems nobody has found.

How to tell whether it worked

The result to measure is rooms released rather than rooms thinned. Track rooms vacated, leases terminated, maintenance agreements and circuits cancelled, cooling and UPS equipment retired, and a restore tested from the new location. Baseline before work starts and measure the same way afterward, including network transport, which can rise once local traffic crosses the wide area network.

Failure modes to watch for:

• A residual handful of servers keeps the old room energized, so the agency pays for both.
• Savings are claimed at cutover while the old contracts still run to term.
• The target platform cannot host what remains: license dongles, serial connections to SCADA or door controllers, unsupported operating systems.
• Latency appears in records, imaging, or dispatch workflows that depended on the server being in the same building.
• Hardcoded addresses and license servers bound to specific hardware surface during a window instead of during mapping.

Working with HQT on this

HQT scopes this kind of work as a fixed-price or time-and-materials engagement with a written statement of work, named staff, and a defined acceptance process. We are candid about what we can take on and prefer a well-run project to an oversized one. Contact us to discuss a requirement, a draft scope, or an upcoming solicitation.

Frequently asked questions