Data center relocation

Data center relocation: planning, risks and implementation

  

What aspects need to be considered when moving a data center?

Moving a data center usually proves to be an enormous challenge. The goal is to get everything back up and running as quickly and reliably as possible. However, the implementation turns out to be anything but easy due to the immense deadline pressure and the many stumbling blocks. In short, the planning and coordination effort is immense.

Typical questions usually arise in the run-up to the event:

  • What preparations need to be made to ensure a smooth process?
  • Is there sufficient documentation available or does it have to be created first?
  • What is important on move days?
  • What happens with the IT: lift & shift of the existing technology or a tech refresh through the use of new hardware?
  • What risks do exist during the actual transport?
  • Will the existing hardware survive the move at all?
  • Are there any spare parts?

The most important decision in planning, however, is: should you go for a "big bang" and tackle everything at once, or does a time-consuming but safe variant with individual phases of moving make more sense?

  

Planning a data center move: "Big Bang" vs. phases of moving

A "Big Bang" refers to a complete move of the affected hardware. Specifically: moving the entire server network landscape 1 to 1 from A to B on a specific date. The second method describes a move in waves, or phases of moving. This means that the IT landscape is moved in various, smaller steps over a longer period of time. Of course, there are pros and cons to each methodology.

"Big Bang": pros and cons

Advantage: Less time-consuming.
Disadvantage: High risk. This is because everything from the documentation of the “old” landscape to the destination landscape must be spot on and completed on the moving day itself. In this case, the move has to be an instant success. In the worst case, problems occur during startup in the new landscape. For example, the network does not work. If there is no time to solve these problems on site, the only solution is to "fall back", i.e. to transport everything back to the old landscape.

Phases of moving: pros and cons

Advantage: Less risk. During the planning phase, you can decide to move the test environment, the production environment, or the storage first, for example, or to determine a completely different sequence. Which of the variants are optimal depends heavily on the size of the data center. In this way, each module can be booted and tested individually. If everything works, then the next part follows. In addition, the learning curve increases with each step in this method.
Disadvantage: Very time-consuming and, above all, demanding in terms of human resources.

Decision support: "Big Bang" or phases of moving?

Whether "Big Bang" or moving in phases are more suitable for a data center move varies from case to case. Our experts always discuss the best method in close consultation with their customers. A decisive factor here is the size of the landscape to be moved. For example, if only three racks are involved, the solution is obvious. If there is a larger amount of hardware, the situation becomes more complicated. This is also the case if various subcontractors are involved in the operation of the data center or, for example, monitoring is controlled from abroad, in which case there are a large number of interdependencies.

  

Preparations for a smooth data center move

Sophisticated planning plays an elementary role, especially in the case of a comprehensive data center relocation. The first step is to draw up a rough phase plan in close cooperation with all those involved. As part of this, the hardware to be moved is documented and thus the asset lists are brought up to date and license and maintenance contracts are checked. The inventory forms the basis for further refinement of the phase plan. For this purpose, it is also necessary to get a picture on site in advance, for example, of thresholds, narrow spaces, floor load-bearing capacity and other obstacles. The final relocation plan then takes into account department-specific conditions and links them to the required phases: dismantling, transportation and reassembly. To ensure that the process is as efficient as possible, those responsible ultimately divide everyone involved into groups.

The special challenge on so-called "move days"

On the day of the move itself, everything has to work as planned. The people who remove the equipment from the data center and move it to the new data center must be present. This may sound trivial, but it requires the best possible coordination. After all, a large number of employees are on standby on move days, and they have to know exactly what to do and when. The challenge is to bring everyone together. A detailed move plan that regulates the cooperation of the various teams helps to ensure that the schedule is adhered to. At the same time, tried-and-tested handover and acceptance procedures enable permanent control of all the necessary work steps.

  

Data center relocation: the role of the IT landscape and its age

As part of the hardware inventory, it is easy to see whether the equipment is already older and therefore perhaps no longer powerful enough or even out of life-time. The customer must then decide whether he wants a so-called lift & shift or a tech refresh. Lift & shift means to dismantle the hardware, such as the rack 1 to 1 - as it is - and set it up at the new location. Another option would be to combine the data center move with a tech refresh. Overall, it's important to find out if new hardware is needed to make the data center perform as well as possible while remaining efficient. Important factors are also new potential maintenance contracts, which ideally require fewer licenses and thus save costs.

Risks in relation to the IT landscape: lift & shift and tech refresh

With a lift & shift, it often looks like the experts are shutting down systems that have been running permanently for five years, for example, for the first time ever during a data center move. This often turns out to be an exciting affair. Because you can never be sure whether these systems can be started up again. One critical point here, for example, is the reliability of hard disks. In the case of a tech refresh, i.e. a move to a new IT landscape, the risk proves to be much lower. In this case, the racks with the new hardware, including all the cabling, are already on stand-by. During the actual data center move, the "old" hardware only has to be switched off and the "new" hardware put into operation. If errors occur, you can "fall back" by simply switching over or switching back.

Minimizing the risks of a lift & shift

Even if it increases the costs ever so slightly: performing tests in advance as part of a lift & shift is highly recommended. Specifically: Simply shut down the systems and then start them up again. Experience shows that this does not always work. You need to be prepared for this scenario and have all potentially required spare parts on hand, specifically new hard disks or power supplies.

  

Data center relocation: transport challenges

It starts with getting there: Due to the tight schedule, the route should be mapped out in advance. Potential disruptive factors such as construction sites, bottlenecks and traffic jams can be identified and planned for. Also important: sleepers and road damage. Although air-suspended vehicles are used to transport the hardware, the bumps must not be too severe. The next step is to take a close look at the "new" data center and walk along all the paths. Thresholds and bottlenecks are also a factor, but here in terms of lift truck suitability. It is also essential that two people can walk past each other in the aisles. This is especially true if there is neither an airlock nor a ramp. After all, a complete rack cannot simply be transported with a pallet truck. Last but not least, the floor must stand up to scrutiny. Most data centers have a raised floor that can only withstand a certain load in terms of kilos per square meter. For example, there have been use cases where companies wanted to move entire racks with them, but half of the hardware had to be removed because, quite simply, the total weight was too high for transport over the raised floor.