By Martin Fröborg, Global Segment Manager – Energy at MCT Brattberg
Backup power and redundancy featured prominently at Data Center World Power in Dallas. This is understandable. Reliable power is essential to the operations a data centre supports. I also left the event with a question that extends across the life of the facility.
How do we know that the protection works today, and will still work after the next change?
A data centre needs to evolve. Equipment is replaced, cables are added, and maintenance is carried out. Each activity must be managed to maintain the intended level of reliability. In my view, the ability to verify this should influence how we design and select technical solutions from the outset.
The consequences call for a broader perspective
Uptime Institute’s Global Data Center Survey 2026 shows that the proportion of organisations experiencing outages continues to fall. Yet 71% of respondents who answered the cost question estimated that their most damaging outage had cost at least US$100,000. This group consisted of organisations that had reported an outage classified as significant or worse within the previous three years. 11
In the same survey, 56% identified power as the primary cause of their most recent impactful outage. [1] This makes backup power a natural focus. But to understand why protective measures fail to work as intended, we also need to understand how installations, procedures, and system dependencies interact.
Addressing human factors before errors occur
Uptime treats human error as a contributing factor that can underlie several types of technical outages. In its 2026 outage analysis, 59% cited failure to follow procedures and 36% cited incorrect procedures as ways human error contributed to outages. Respondents could select multiple answers to this question about human factors; the figures are not a breakdown of all outages.2
This raises questions about the conditions we create for the people doing the work. Is it clear what a correct installation looks like? Can changes be identified? Can the next technician understand the installation without finding the person who built it?
Training and clear procedures matter. I believe you should also consider the ability to inspect and verify a solution when selecting it. Instructions become difficult to follow if components cannot be identified or inspection points cannot be accessed.
In operations where errors can have serious consequences, safety should depend as little as possible on individual memory and judgment. Making it possible to do the work correctly, and to confirm that it has been done correctly, should therefore be part of the same design task.
Redundancy needs to be tested as a system
Google Cloud’s incident in Paris in April 2023 illustrates how several weaknesses can interact. A leaking cooling water pipe elsewhere in the facility allowed water to reach a UPS room, where a fire broke out. When the building’s power was shut off, an incorrect distribution of database replicas across buildings contributed to the regional impact. A smoke-damaged server cluster was decommissioned after its data was recovered.3
During Cloudflare’s outage in Oregon in November that year, dependencies emerged that earlier tests had missed. The company had tested parts of its redundancy but not the complete loss of the affected facility. Some services became unavailable, while traffic continued to flow through Cloudflare’s network.4
My takeaway from these incidents is that verification needs to cover how systems actually depend on one another. When we make a change, we should ask whether previous testing still covers the new configuration and any additional dependencies.
Physical protection must remain effective
In Dallas, I also spoke with representatives from FM. Its data centre guidelines provide a clear loss prevention perspective. Smoke and liquids are identified as key risks to sensitive electronics. The guidance also addresses the positioning of power equipment and cabling to prevent a fire from defeating redundancy.5
Cable and pipe transits show how this becomes a practical design issue. FM recommends that penetration seals in fire-rated walls and floors match the fire resistance of the surrounding construction and meet specified leakage requirements. [5]
For ongoing operations, the question is how that protection is maintained when an installation is modified. The required checks and who is responsible for carrying them out must be clear before work begins.
Checking the existing installation when a transit is opened
The ability to add new cables to a transit is important. It is at least as important to check the cables and sealing blocks already in place when the transit is opened for a modification. The next installation should start from a verified condition.
The dimension markings on MCT Brattberg’s blocks support this in practice. The block size can be read from the outside, while the marking for the intended cable diameter is on the inside.
The marking can then be compared with the cable’s measured outer diameter and the manufacturer’s sizing table to check that the correct block has been selected.6 This gives the technician a physical reference to check against, even when someone else carried out the installation. It reduces the need to assume the existing arrangement. Checking dimensions is one part of inspecting the transit; the remaining installation details and reassembly also need to be verified against the installation instructions.
Inspectability starts at the design stage
I would like to see this perspective feature more clearly in discussions between clients, designers, suppliers and operators. When a solution is specified, it should also be clear how the installation can be checked throughout the facility’s life and during future modifications.
At handover, the documentation needs to describe the installation as built and the checks carried out. After work on the installation, restoring protective measures, updating documentation and verifying the affected functions should form part of completing the job. Whoever takes over should be able to understand the facility’s current condition.
For me, Dallas was a reminder that reliability needs to be maintained as a facility evolves. The decisions we make during design shape the conditions for work that will be carried out much later.
The next project review should therefore include a simple question: How do we check what is installed today, and demonstrate that the protection still works after the next modification?
Article sources:
- Uptime Institute Global Data Center Survey 2026
Pages 19–22, particularly Figure 13 on costs (n = 77) and Figure 14 on outage causes (n = 85). The cost group had experienced Category 3–5 outages within the previous three years. Costs are respondents’ estimates from the start of the outage to full recovery, including direct and indirect costs. ↩︎ - Uptime Institute Annual outage analysis 2026
May 2026. Pages 19–21, particularly Figure 14 on human factors. Respondents could select up to three answers; n = 199. The statistics cover IT service outages and are not limited to physical installations. ↩︎ - Google Cloud Incident Report for europe-west9
Final report published June 23, 2023. The incident began on April 25 in the US time zone used by the report, or April 26 local time in Paris. Sections Water Leak and Fire Damage, Regional Impact, and Water Leak, Mitigation, and Prevention. ↩︎ - Cloudflare Post mortem on the Cloudflare Control Plane and Analytics Outage
November 2023. Sections Intended Design and Designing for Data Center Level Failure. The example in this article draws on Cloudflare’s account of its own system dependencies and testing. ↩︎ - FM Data Sheet 5-32 Data Centers and Related Facilities
January 2026, revised July 2026. Sections 1.2 on hazards, 2.2.6 on penetrations and 2.6.3 on power and utility cabling. FM’s recommendations are presented as loss prevention guidance. ↩︎ - MCT Brattberg Standard Block ↩︎
Original announcement: MCT Brattberg
Image: Courtesy of MCT Brattberg; Stock or AI images for illustration purposes only
Explore more about MCT Brattberg
About this Advertorial Feature
This article has been published as an Advertorial Feature as part of Cable Technology News’ annual Industry Partner Programme.
Advertorial Features are professionally presented and published in accordance with our Editorial & Advertising Policy. Publication does not constitute an endorsement of any company, product or service.
Interested in raising your organisation’s profile across the global cable industry?
Explore our annual industry partnerships here.
