Data Center Maintenance · Case study
Colocation Hardware Refresh Across Frankfurt, Amsterdam and Rotterdam.
A representative walkthrough of a multi-site colocation refresh: new racks, clean cabling and audited decommissioning.
Representative scenario. Last reviewed October 2026.
Engagement profile
- Customer
- Software and managed services provider with leased cabinets in three colocation sites
- Scale
- 3 colocation sites, about 24 cabinets, roughly 160 devices to install and 140 to retire
- Service
- Data Center Maintenance
- Where
- GermanyNetherlands
Representative scenario. This case study shows how Deploris scopes and delivers this kind of work. It does not describe a real client, and no results are claimed. Want a plan like this for your own sites? Request a quote.
Quick answer
What does this engagement look like?
A colocation hardware refresh in Frankfurt, Amsterdam and Rotterdam means racking new servers and switches, re-cabling to a labelled standard, and decommissioning legacy equipment. Deploris runs it under one engagement manager, with written scope and approved methods of procedure executed by vetted local engineers. The customer receives cable run sheets, photos, test results and chain-of-custody records.
THE CHALLENGE
The challenge.
The customer leases cabinets in one Frankfurt campus and two Dutch facilities, and the hardware in them spans several generations. Older servers sit beside newer ones, patching has grown by accretion, and the rack elevations in the DCIM tool no longer match what is physically installed. The refresh adds new compute and top-of-rack switching, then retires the legacy estate once workloads have moved. The risk is not the racking itself. It is that nobody can say with confidence which cable feeds which port, which power supply sits on which PDU feed, or which old server still carries a dependency.
The work also crosses borders and operators. Each facility has its own access process, dock rules, cross-connect ordering and out-of-hours policy, and the customer's own team is remote and only wants to cut workloads over once the physical layer is verified. Legacy drives may hold personal data, so disposal needs documented handling, not a pallet collection. The engagement therefore has to behave like a single controlled change program, even though it is delivered by different local engineers in different buildings.
Constraints the plan had to respect
- Each colocation operator requires named visitors and advance access requests, with photo ID checked on arrival, so engineer rosters are fixed before the change window opens.
- New equipment arrives from outside the EU for part of the order, so customs clearance and delivery appointments at each loading dock set the earliest possible install date.
- Legacy equipment stays powered and cabled until the customer's application owners confirm workloads have moved, so new and old hardware must coexist in the same cabinets and power budgets.
- Working-time rules in Germany and the Netherlands limit shift length and require rest periods, so long overnight windows are staffed in relays rather than by one engineer.
- Drives and other data-bearing parts must be sanitized or destroyed with a documented record, because they may contain personal data covered by the customer's GDPR obligations.
HOW THE WORK RUNS
How the work is run.
- 1
Discovery and rack baseline
The engagement manager starts with the customer's DCIM or NetBox export, the purchase order and the colocation contracts, then sends a local engineer to each site for a baseline walk. The engineer photographs every cabinet front and rear, scans asset tags, records PDU feed, outlet and phase for each power supply, and notes cable types and existing label schemes. Differences between the export and the physical racks are logged as findings. The result is a verified starting point, so the later decommission list rests on what is actually installed, not on what the spreadsheet says.
- 2
Written scope, methods of procedure and access
Before any hardware moves, Deploris issues a written scope and one method of procedure per site. Each MOP lists target U positions, power budget per cabinet, cable run sheet, fiber and copper types, labelling convention (TIA-606 style or the customer's own), test steps, rollback and a stop point. Change records are filed in the customer's ITSM tool. The manager books loading dock slots, requests access for named engineers with each operator, and places cross-connect orders using the customer's letters of authorization. Frankfurt documents can be bilingual where the site requires German paperwork.
- 3
Receiving, inspection and staging
Shipments are pre-notified to each facility with PO numbers and consignee details, and a local engineer receives them at the dock. Each device is checked against the packing list and PO by serial number, inspected for damage and photographed. Rails are pre-assembled and labels are printed ahead of install, so cabinet time is spent installing, not preparing. Packaging is removed to the operator's designated area, because cardboard is normally not permitted in the data hall. Any shortage or damage is reported to the engagement manager before the window opens, not discovered during it.
- 4
Install in waves with the legacy estate left running
Engineers rack and stack the new equipment in waves, cabinet by cabinet, following the approved elevations. Power goes to separate A and B feeds with outlet and phase recorded, copper is dressed through managers to the labelled run sheet, and fiber is cleaned, inspected and tested with an optical power meter before patching. Each device is powered on, checked at POST and confirmed on its out-of-band management port (iDRAC, iLO or equivalent) so the customer's team can image and configure it remotely. Legacy gear stays untouched until the customer signs off each workload move in writing.
- 5
Audited decommissioning and closeout
Once application owners confirm a legacy device is idle, engineers follow a defined sequence: power down, hold for an agreed quiet period, disconnect, remove, then tag and record. Drives are removed or the device is escalated for sanitization, and a work order captures serial numbers, asset tags and destination. A chain-of-custody record follows the equipment to a certified recycler, with a destruction or erasure certificate returned for data-bearing parts. The closeout report then assembles cable run sheets, before and after photos, test results, updated elevations and CMDB reconciliation. After handover, 24/7 remote hands remain available for follow-up work.
WHAT WAS PRODUCED
What the engagement delivers.
Each item links to a full guide with its scope, process and FAQs.
WHAT THE CUSTOMER ENDS UP WITH
What the customer ends up with.
- 01A verified rack baseline for every cabinet, so the customer's DCIM or NetBox records match what is physically installed.
- 02New servers and switches installed to approved elevations, with power feed, outlet and phase recorded for each power supply.
- 03A labelled cable run sheet and photographs for every new run, so any port can be traced from either end without opening a cable bundle.
- 04Legacy equipment retired against signed work orders, with serial numbers, asset tags and chain-of-custody records for each device.
- 05A structured closeout report with photos and test results that the customer's auditors and operations team can file as evidence.
WHAT APPLIES TO YOUR PROJECT
Decisions that apply to similar projects.
Audit before you plan the move. A physical baseline catches drift between the DCIM export and the racks while it is still cheap to fix, and it prevents the wrong server from appearing on a decommission list.
Keep legacy gear running until the application owner signs off in writing. Installing new hardware first lets the customer cut over on their own schedule, and it keeps a rollback path open during the quiet period.
Treat disposal as part of the install plan, not an afterthought. Deciding in advance who removes drives, which certificate comes back and which recycler takes the equipment avoids stranded pallets and data-handling questions after the window closes.
WHERE IT APPLIES
Metros in this scenario.
Local engineers in each metro, coordinated by one engagement manager. Open a metro to see local details.
QUESTIONS
Frequently asked questions.
How long does a colocation hardware refresh take to plan?
Can local engineers work in Frankfurt and Amsterdam data centers outside business hours?
What happens to data on old drives during decommissioning?
More case studies
Other scenarios we have planned.
QUOTE
Plan a project like this.
Share a few details and we will come back with the next steps.
Get started
Need engineers on site? Tell us where.
Tell us the sites, the scope and the timeline. We reply with a written plan.