How to Choose a Rack and Stack Provider for a UK Data Centre

Choosing who racks your hardware is awkward because the work is hard to judge in advance. A neat rack and a messy rack look the same on a quote. The difference appears months later, when something fails, when you add servers, or when a new engineer opens the cabinet and has to work out what is connected to what.

This guide sets out what to check before you appoint a rack and stack provider in a UK data centre, the questions to put to them, and how to decide between the colocation operator’s engineers and an independent provider. If you want to prepare your own side first, our rack and stack checklist covers what to have ready.

1. Facility access and coverage

A provider can only work where they can get in. Facilities control who is allowed on the floor, so this is the first practical question.

• Which sites do they already attend? Familiarity with a building means they know its access process, working rules and how cross connects are ordered. On our cabinet build case study, prior experience of the site meant its policies and procedures did not need to be worked out on the day.
• How long does approval take for a new site? Ask what they need from you and from the facility.
• Where are they based, and how far will they travel? A London hub is useful for London facilities. For regional sites, ask who attends and how it is scheduled.
• What are their working hours? Ask whether out-of-hours work is available and how it is arranged. A provider should not promise round-the-clock cover they cannot show.

DACPROS works from a hub in Hammersmith, London, covering London and the Slough corridor, and attends sites across the UK.

2. Experience with your hardware

Servers from different vendors have different rails, airflow, management controllers and quirks. Generic experience is not the same as experience with your kit.

• Ask which vendors and platforms they have installed recently. Dell, HPE, Lenovo, Cisco, Juniper, Arista, Nvidia and so on.
• Ask about unusual hardware. High-density GPU servers can be long and heavy and need planning for depth, weight, power and cooling.
• Ask who configures it. Racking is one skill. Setting up switches, remote management controllers and operating systems is another. Find out whether the person on site can do both or whether another engineer is involved.

In our cabinet build case study, Dell servers needed iDRAC set up and Linux installed, and several carried network cards that needed a custom driver compiled before they worked. The racking was the smaller half of the job.

3. Installation methodology

Ask how they work, not only what they will do. A provider with a repeatable method will describe it without hesitation.

• Do they review the plan first? Design review before ordering catches problems such as switch airflow, depth and missing cables when changes are still free.
• How do they check the order? Reconciling the order against the design finds what is missing before the engineer is on site.
• What do they do on arrival? Unpacking, serial number capture and checking delivered kit against the inventory.
• How is the work sequenced? For larger jobs, the sequence determines the schedule. On our data centre migration, an advance team prepared the new cabinets, racking switches and laying fibre cross connections, instead of waiting for the old cabinets to be decommissioned.
• Who does what? A sensible provider puts the right level of engineer on each task, so you do not pay senior rates for routine mounting.

4. Cabling and labelling standards

Cabling quality is easy to see once you know what to look for, and difficult to fix later.

• What standard do they work to? Look for dressed, colour-coded runs, correct bend radius and labels at both ends. Our structured cabling best practices is the standard we hold ourselves to.
• What if you have no convention? A good provider can apply theirs and document it.
• How are labels checked? The test is whether every label matches the port map.
• Do they check lengths before the day? Cabling that reached one position does not always reach another, so this matters when equipment is being moved or replaced. See our border router case study.

On our live core cabinet rebuild, every connection was labelled to the documentation, which is what kept troubleshooting at handback to a minimum.

5. Testing requirements

“Tested” can mean very different things. Ask what is tested and how it is recorded.

• What is tested? Power, hardware status, link state and speed, copper and fibre runs, management access.
• Against what? Tests should check against your port map and acceptance criteria, not just whether a light is on.
• How are results recorded? Test results should be in the handover pack, not only in the engineer’s head.
• What happens when a test fails? Ask who investigates and how the fault is recorded.
• Who signs off? Agree what “accepted” means before the work starts.

6. Documentation and photographic evidence

Documentation is the main thing you are left with when the engineers leave.

Ask to see an example handover pack. It should normally contain:

• an as-built elevation and port map
• an asset list with serial numbers
• photographs of each cabinet, front and rear
• test results
• a record of any faults and open items
• configuration and addressing records, where configuration is in scope

Photographs matter for remote teams. If nobody from your organisation is on site, photographs and as-built records are your evidence that the work was done as specified. DACPROS builds are handed over with a photographed pack and as-built records.

7. Change control

Data centre work often touches live systems and shared space. How a provider handles change tells you how they handle risk.

• Do they agree a plan and window before attending? Work on live equipment should have an agreed sequence, window and rollback.
• How are changes to scope handled? If the plan changes on site, who approves it and how is it recorded?
• How do they work with your change process? If you run formal change management, ask how they fit in.
• What do they do if the work cannot be completed in the window? A provider should say how they will leave the environment safe.

Our border router case study is a useful example. The window was proposed by DACPROS and the work completed inside it, with 30 fibre connections moved and no rollback.

8. Escalation process

Something will not go to plan on at least some jobs. What matters is how it is handled.

• Who is the named point of contact? One owner for the job, reachable during the work.
• What happens when the engineer on site cannot resolve something? Ask whether a more senior engineer is available and how quickly.
• How are you told? Ask how and when you hear about problems, not only successes.
• What is the route after the visit? If something fails after handover, ask how you report it and who responds.

Questions to ask before you appoint a provider

Put the same questions to each provider and compare the answers.

1. Which UK data centres do you already attend, and what is the approval process for a new one?
2. What are your working hours, and is out-of-hours work available by arrangement?
3. Which hardware vendors and platforms have you installed recently?
4. Who configures switches, management controllers and operating systems, and are they on site?
5. Do you review the design and check the order before the hardware arrives?
6. What cabling and labelling standard do you work to, and can you apply ours?
7. How do you check cable lengths and port mapping before the day?
8. What do you test, and how are the results recorded?
9. Can I see a sample handover pack?
10. What photographs do you take, and when do I receive them?
11. How do you agree a change window and handle scope changes on site?
12. Who is my point of contact, and what is your escalation route?
13. What happens if the job cannot be finished in the window?
14. Can you give references or case studies for similar work?

Colocation operator or independent engineer?

Many colocation operators offer on-site assistance, and some can rack equipment. Whether that is enough depends on what you need. Services, hours and charges vary by operator, so read the terms.

The operator’s engineers are often a good fit when:

• the job is small, such as a few devices
• you can write exact, step-by-step instructions
• you do not need configuration, a formal test record or detailed documentation
• speed and convenience matter more than a documented build

An independent engineer is often a better fit when:

• the job is a planned build with a lot of cabling
• it includes switch, controller or OS configuration
• you need a design review, a labelling standard and a handover pack
• your team is remote and needs evidence the work was done as specified
• you use several facilities and want one consistent standard across them
• you want the same provider to return for later growth and support

The two are not exclusive. Some customers use the operator for quick instruction-led tasks and an independent provider for builds. Our guide to remote hands vs smart hands explains the difference between instruction-led and judgement-led work, which is a useful test for which route to take.

What to be wary of

• Vague answers on testing or documentation. If they cannot describe a handover pack, you may not get one.
• Quotes with no written scope. A price that does not say what is included cannot be compared.
• No design review. Providers who simply turn up and rack what arrives leave the planning risk with you.
• Claims of universal coverage. Ask which sites they have worked in.

How DACPROS answers these questions

DACPROS provides rack and stack as part of its build and deployment service. We work from a hub in Hammersmith, London, and attend sites across the UK. Builds are planned around your access and change windows, tested and handed over with a photographed pack and as-built records. DACPROS lists CompTIA A+, CDCMP, PRINCE2 and ITIL among its engineer accreditations.

For worked examples, see the two-rack cabinet build, the seven-cabinet data centre migration and the live core cabinet rebuild. To prepare your side of the project, use the rack and stack checklist.