Server Relocation Services: Checklist, Downtime & Cost (2026)

Short answer: A server relocation is a change-window project, not a removals job. Plan it as inventory, dependency map, backup verification, shutdown order, transport, rebuild and test, with one maintenance window per rack or group of systems. How much downtime you need depends on whether you do a cold move, a staged move or a lift-and-shift to cloud. Cost is driven by rack count, distance, after-hours windows, re-cabling and data centre access, not by weight.

When does a server relocation happen?

Most server moves in Hong Kong and Singapore come from one of four triggers:

• An office move. The server room or a few racks in the old office need to go to the new one, or to a data centre instead.
• A colocation exit or change. A lease ends, a contract is renegotiated or a facility is being closed, and the racks must go to another site.
• A consolidation. Equipment from several rooms or offices is brought into one data centre, or two data centres are merged.
• A refresh or a cloud decision. Some of the equipment is retired, some is moved and some workloads are rebuilt in the cloud.

Brocent's IT relocation services cover server and data centre moves as well as office IT moves, from un-rack and transport to re-rack, testing and post-move support. This guide sets out what a sound plan looks like so you can judge any provider's proposal, including ours.

What does the discovery phase need to produce?

The quality of a move is decided before anything is switched off. The discovery phase should end with four documents:

• An asset inventory listing every server, storage array, switch, firewall, UPS and cable, with asset tags, serial numbers, rack positions and ownership. A server on a leased contract may belong to the vendor.
• A dependency map showing which applications run on which servers, which depend on which databases, and which talk to the outside world. This decides the shutdown and start-up order.
• A priority list that ranks systems by business impact, so the most critical are brought back first.
• A risk and rollback plan that states what happens if the move fails at each stage, and who decides to roll back.

Ask who confirms the inventory. Brocent's process asks the client to confirm the IT asset list and verifies its completeness before work starts, because a missing server discovered on move night is an expensive kind of surprise.

What does the server relocation checklist look like?

Use a countdown checklist so that every task has an owner and a deadline. The timings below are a template to adapt, not a standard.

T-30 days: scope and booking

• Confirm the destination: power, cooling, rack space, network and cross-connects.
• Book the data centre access and any loading dock slot at both ends.
• Agree the move windows with the business, usually evenings or weekends.
• Confirm insurance, vendor warranty and support terms for the move.

T-7 days: preparation

• Verify backups and run a restore test on the most critical systems.
• Lower DNS time-to-live values so address changes take effect quickly.
• Label cables and ports, photograph racks, and update the rack diagram.
• Confirm network configuration, IP plan and firewall rules for the new site.
• Brief the people who will test applications after the move.

T-1 day: final checks

• Take final backups and confirm they are complete and stored away from the equipment being moved.
• Confirm the shutdown order and who gives the go-ahead.
• Check that packing materials, tracking and transport are in place.

Move night

• Shut systems down in the agreed order and confirm each is powered off cleanly.
• Un-rack, pack with an itemised pack slip and check each asset tag against the inventory.
• Transport with supervised loading and unloading.
• Rack, cable and power on in sequence at the destination.

T+1 day: validation

• Run the test cases, compare the asset list with what arrived and review the network and application tests.
• Keep the old site available for a rollback window.
• Record issues and close the project with a signed-off report.

How much downtime should you plan?

Downtime depends on the migration style, not on the distance. There are three common styles.

Cold move vs staged migration vs move to cloud

• Cold move. Everything is shut down, moved and restarted in one window. It is the simplest to plan and has the longest outage, usually a night or a weekend. It suits systems that can be offline outside business hours.
• Staged migration. Systems move in groups over several windows, or a temporary copy is built at the destination and traffic is switched over. Downtime per system is short, but the plan is more complex and you run two sites for a while.
• Move to cloud. Some or all workloads are rebuilt in the cloud instead of being moved. Downtime can be minimal, but it is a migration project with its own testing, licensing and cost questions, not a transport job.

Many firms combine styles: cold-move the simple, non-critical systems and use a staged approach for the few that cannot go offline. Ask your provider to state the expected outage for each system in minutes or hours, not in general terms.

What about transport, insurance and chain of custody?

Equipment in transit is exposed to shock, theft and loss, so treat transport as a controlled process. Brocent's process includes GPS trackers on equipment, packing with an itemised pack slip, special packaging for non-standard devices, supervised loading and unloading, and checking that insurance is valid before the move. Whichever provider you use, ask for the same things in writing:

• Who insures the equipment, to what value, and whether the cover applies door to door.
• How assets are tracked from rack to rack, with signed hand-offs.
• How non-standard devices are protected, such as tape libraries, storage arrays and rack-mounted UPS units.
• What happens to a damaged item, including who supplies a replacement and who covers the downtime.

What drives the cost of a server relocation?

Providers price server moves differently, so ask for a quote that shows its components. The main cost drivers are:

• Number of racks and devices. More equipment means more un-rack, pack, rack and test time.
• Distance and route. Moves within one building cost less than moves across the city, and cross-border moves add export and import handling.
• Time windows. Evening, weekend and holiday windows cost more than business-hours work.
• Re-cabling and network work. A destination that needs new cabling, cross-connects or a changed IP plan adds scope.
• Data centre fees. Access requests, escorts, loading dock bookings and remote-hands charges at either end.
• Project management and testing. The planning, the move-night supervision and the post-move test cycle.
• Rollback readiness. Holding the old site open for a rollback period has a cost.

Brocent does not publish a relocation price list, because scope varies too much. Ask for a quote built from these components for your racks, and compare quotes line by line.

What are the access rules in Hong Kong and Singapore data centres?

Data centres in both cities generally control who can enter, when equipment can arrive and how it is handled. Typical requirements include advance notice and named visitors, an escort or badge process, a booked loading dock slot, inspection of incoming equipment, and rules on packaging materials in equipment rooms. The details differ by operator and by facility, so confirm them with each site before you set the move date. If the move crosses the Hong Kong and Singapore border, add export and import paperwork to the plan and allow extra time.

When do you need the vendor's engineers on site?

Some equipment, especially storage arrays, mainframes and specialised network devices, is covered by support contracts that require the vendor or a certified partner to un-rack or re-commission it. Check each contract before the move. If vendor attendance is needed, book it with the same lead time as the data centre slot. Where the equipment is multi-vendor, a relocation provider can coordinate the engineers so the right people are on site at the right time. For the data centre side of the project, read the Hong Kong data centre relocation guide and the account of a rack migration between data centres.

How do you handle network and IP changes in a move?

If the servers keep the same addresses at the destination, the move is much simpler, but that is rarely possible when the site changes. Decide early which of these applies:

• Same IP plan, new site. The destination network is built to match, so servers and clients need no changes. This usually needs a stretched network or a carefully prepared cut-over.
• New IP plan. Every server address, firewall rule, DNS record, certificate binding and application setting that mentions an old address must be found and changed. The dependency map is what tells you where to look.
• New internet and WAN links. Lead times for circuits and cross-connects are often longer than the move itself, so order them first and track the delivery date.

Lower DNS time-to-live values a week before the move, keep a list of firewall changes ready to apply, and confirm that licences tied to IP addresses or hardware identifiers will be accepted at the new site. Brocent's relocation process includes a network requirements assessment at the site evaluation stage, because network surprises are a common reason a server comes up in the new room and still cannot be reached.

How do you test after the move?

Powering a server on is not the same as the service working. Agree the test plan in advance, and sign it off with the application owners.

• Infrastructure tests. Power, cooling, network connectivity, storage visibility and time synchronisation.
• System tests. Each server boots, services start, and disks and databases mount in the expected order.
• Application tests. Business users run a short list of real tasks, such as logging in, running a report and posting a transaction.
• Integration tests. Links to other systems, printers, backups, monitoring and remote access.
• Backup and monitoring. The first backup at the new site completes, and monitoring alerts reach the right people.

Only after these pass should the old site be released. Brocent's process ends with a validation and user acceptance stage that reviews the updated asset inventory, the test reports and the time and material records, so the project closes on evidence rather than assumption.

What does a rollback plan look like?

A rollback plan answers the question every board member asks: what if it does not work? It should state the point of no return, who decides, and how to go back.

• Define the decision points. For example, after the shutdown, after the first power-on at the new site, and after the application tests.
• Keep the old site intact until the tests have passed and the rollback window has closed. Do not decommission or hand back the old space on move night.
• Keep data consistent. If systems have run at the new site and then must go back, plan how changes made there are carried back.
• Name the decision maker and make sure they are reachable throughout the window.

What are the common reasons a server move goes wrong?

Many failures are planning failures rather than technical ones. The patterns below recur.

• An incomplete inventory. A forgotten server, appliance or cable is found only when something breaks.
• Hidden dependencies. An application depends on a server nobody listed, or on a hard-coded address.
• Untested backups. The backup exists but has never been restored, so there is no safety net.
• Underestimated lead times. Circuits, cross-connects, access approvals and vendor engineers take longer than the move.
• No clear owner on move night. Nobody can decide whether to continue or roll back.
• Packing damage. Equipment is moved without proper packaging, tracking or insurance.

Case study: a trading firm's server room

*This is an illustrative example, not a named client.* A 120-person Hong Kong firm was leaving an old office and had one server room of six racks. It built an inventory and dependency map, classified its systems into three groups, and cold-moved the file and print servers on a Saturday night while building a temporary copy of its trading-support database at the new site for a staged switch-over. Each rack had its own checklist, and the old room stayed available for a week as a rollback option.

Frequently asked questions

How long does a server relocation take?

The move itself often takes one night or one weekend per group of racks, but the planning takes weeks. Expect discovery, booking and preparation to be most of the elapsed time.

How much downtime should we expect?

For a cold move, the outage is the length of the shutdown, transport, rebuild and test window, commonly a night or a weekend. A staged migration can keep the outage for each system to minutes or hours. Ask for the expected outage per system.

Who insures the equipment during the move?

It depends on the contract. Ask your provider to confirm in writing who insures the equipment, the insured value and whether cover applies from the old rack to the new one.

Do we need the vendor's engineers on site?

Only where your support contract requires it, which is common for storage arrays and specialised devices. Check each contract and book the engineers early.

Can we move to cloud instead of moving the servers?

Yes. Moving some or all workloads to the cloud avoids the physical move, but it is a migration project with testing, licensing and cost questions. Many firms use a mix of moving and rebuilding.

What does a relocation quote include?

A good quote lists discovery, project management, un-rack and pack, transport and insurance, rack and cable at the destination, testing and post-move support, and states the assumptions on time windows and data centre fees.

How early should we start planning a server relocation?

Start as soon as the move date is known, and allow several weeks to a few months ahead for a multi-rack move. The long-lead items are network circuits, data centre access approvals, vendor engineers and the discovery work that produces the inventory and dependency map. A short timeline is possible for a small move, but it leaves less room for testing and rollback.

Should we move old servers or replace them?

Compare the cost and risk of moving each system with the cost of replacing it. Equipment that is out of support, near end of life or has a poor failure history is often better retired and rebuilt, because a move is the moment a weak component is most likely to fail. Decide this during discovery, before the move date is fixed.

Can we keep running while the servers move?

Only if the systems you need are duplicated or moved in stages. In a cold move, anything on the servers being moved is offline for the window. If a service must stay up, build a temporary copy at the destination, switch users over, and then move the original, or move that service to the cloud first.

Next steps

Read how the process works on the IT relocation services page, see how infrastructure builds are delivered in IT infrastructure deployment services, and for the equipment side of a data centre exit read colocation in Hong Kong with remote hands. For a scoped quote for your racks and dates, talk to a consultant.