Data center emergency: back in operation within 48 hours

The first 60 minutes

The first hour after a failure sets the course for the entire recovery. Experience shows that most of the time is lost not through missing technology, but through unclear responsibilities and people acting in parallel on their own.

1. Protect people. In the event of fire, smoke or water ingress the rule is: clear the room, trigger the emergency response chain, no rescue attempts for hardware. Servers are replaceable.
2. Narrow down the cause. Is power, cooling, network or the building envelope affected? That answer decides whether replacement infrastructure can be operated at the same site at all.
3. Shut down in a controlled way. If the cooling fails, depending on room size and occupancy there are often only a few minutes before critical temperatures are reached. Shut down systems that are not strictly needed in an orderly manner – this prevents hardware damage and data inconsistencies.
4. Set up a crisis team. One person decides, one person communicates internally and externally, one person documents times and actions. The documentation is later relevant for insurers and auditors.
5. Request replacement capacity in parallel. The most common and most expensive mistake: waiting for the loss adjuster's report before sourcing a replacement. Enquiring in parallel costs you nothing – waiting costs days.

Which failures call for a replacement solution

Not every incident creates a need for replacement capacity. The following classification helps you judge whether a mobile solution makes sense.

The 48-hour sequence in detail

The sequence below describes the standard case with a standard system available and a suitable site. It is not a guarantee – but it is a realistic benchmark for what is organisationally possible.

1. Hours 0–2 Initial call and feasibility

   Capturing IT load, site, available connected load and target date by phone. The result is a clear statement of whether delivery is possible and in what configuration.

2. Hours 2–6 Configuration, quote, approval

   Defining cooling, UPS, a backup generator if required, and connectivity. In parallel, access routes and the crane standing area are checked – ideally from photos or a satellite image of the site.

3. Hours 6–24 Preparation and transport

   The system is prepared for deployment, loaded and driven to the site. For larger systems, special permits and the crane are organised in parallel.

4. Hours 24–36 Positioning and connection

   Setting down on the prepared footprint, alignment, connection to power and network, commissioning of cooling, UPS and fire protection.

5. Hours 36–48 Function testing and handover

   Testing of all systems, connection to remote monitoring, handover to your team. From this point on, racks can be populated and systems started up.

Where time gets lost

In practice the critical path almost never runs through the data center itself, but through site preparation. These three points cause most of the delays:

The power connection

If there is not enough connected load available at the replacement site, either the supply cable has to be upgraded or a backup generator has to be planned in. Upgrading involves the network operator and normally takes weeks. A backup generator is more expensive to run, but immediately available – and it makes the schedule independent of network reinforcement work.

The footprint

What is needed is a sufficiently level surface that carries the load, plus access for lorry and crane. Lawns, underground car park decks and tight courtyards are the classic problem cases. A photo of the intended area and the access route answers more questions in the first call than any written description.

Internal decision paths

Approvals, procurement rules and insurance queries regularly cost more time in an emergency than transport and installation combined. Defining an escalation and approval procedure for emergencies in advance saves exactly those hours.

The 5 details that speed everything up

With this information, a reliable feasibility statement is usually possible within a few hours:

• IT load required in kW. What absolutely has to keep running? A fraction of the original capacity is often enough at first to cover the business-critical systems.
• Available connected load. What capacity is actually available at the replacement site, and how far away is the transfer point?
• Footprint and access. Address, ground surface, approximate dimensions, access width – ideally with a photo.
• Network connectivity. Is fibre usable on site, or is a link via 5G/LTE or microwave radio needed?
• Target date. From when do the systems have to be running again? This determines which compromises in the configuration make sense.

Being prepared before it happens

The most effective lever for short recovery times sits before the event. Three measures with a very good effort-to-benefit ratio:

• Record a site profile. Document the connected load, possible footprints, the access situation and contacts once. That removes the entire clarification phase when it matters.
• Define a minimum configuration. Which systems really have to run first? Settle it in advance and you will not be debating it under time pressure.
• Clarify the approval route for emergencies. Who may place an order in an emergency, and up to what amount, without further sign-off?

How the costs of such a replacement solution are made up is covered in the article Renting a mobile data center: the factors that determine the price.

Frequently asked questions