RZ refurbishment during ongoing operation

Server generations come and go. As a rule of thumb, IT equipment is replaced every three years. What remains is the data center and its infrastructure. Typically every 15 to 20 years, there is an opportunity for major adjustments. This is usually triggered by aging infrastructure components or by changes in the requirements of the next server generation (which is changing once again). More performance in less space leads to more power density, but also places increased demands on power supply, cooling and monitoring in particular.

Higher availability

Whereas IT used to be a "means to an end" and was used more as a supporting tool, nothing works without IT today. Even short interruptions lead to work stoppages and additional expenses - a half-day Internet outage is often considered a minor catastrophe in a large corporation. But smaller companies are also experiencing the change. Whereas a power outage at the company headquarters used to be accepted, there is now an expectation, particularly in the case of remote offices, that normal work can be guaranteed despite a local power outage.

Time for change

Whatever the trigger, changes in the data center cost time and money. Meanwhile, data center space is one of the most expensive building spaces of all. Construction costs can quickly reach CHF 15,000 to CHF 25,000 per square meter. It is worth asking a few basic questions before investing:

• Location risks: Is the data center in the right place?
• Capacity: What space and services are needed?
• Availability: What downtime per year is tolerated?

Energy efficiency and green IT are nice buzzwords, but experience shows that they are mainly tackled when it is also financially worthwhile. Major conversions or new buildings are precisely such opportunities to take a step forward here as well. The PUEDA support program is carried out as part of the competitive bidding process to increase the electricity efficiency of the federal government and can create additional financial incentives.

Green meadow in the basement

Who wouldn't want to leave all the legacy behind and redesign a data center from the ground up? Recent examples in the Swiss data center industry show what innovative solutions are possible. For many data center operators, however, it usually means coming to terms with the existing location after weighing up the options of new construction, co-location and renovation. The "green field" is then just in the own basement, not green, but already preloaded. It still has to be operated, especially during a rebuild.

Refurbishment during operation

If an existing data center is structurally renovated, this brings with it undesirable side effects. Dust, dirt, vibrations, non-IT personnel and interventions in the vital building services are the nightmare of every data center operator. If entire walls are torn down, new cooling lines are welded into place and the main electrical distribution is replaced, then it becomes difficult to guarantee fail-safe operation, data protection and data security during this time. "Escape" seems to be a good remedy here.

Provisional RZ

Temporary relocation of IT is a good option for larger data center conversions. Ideally, suitable rooms are already available in the immediate vicinity. It is worth inquiring about the corresponding capacities within the company, with co-location providers or with neighboring companies. If there is no suitable data center space available, it may be worthwhile to set it up temporarily.

"Turnkey data centers can be obtained from various suppliers in container construction. For (outdoor) installation, space as well as power and data connections are required first and foremost. If required, emergency power supply and cooling production can also be provided in containerized form. Modular data centers of this type (temporary and fixed) are particularly common in the USA. In this country, the trend has not (yet) arrived. However, there are arguments in favor of temporary solutions, whereby the rental costs usually exceed the purchase price after ten to twelve months.

Empty basement rooms are also suitable for temporary data centers. Shelters in particular can serve as a suitable solution during the construction phase with appropriate conversion. It is important to ensure a clean separation from the construction perimeter so that the existing data center can be converted as unencumbered as possible after the temporary relocation of the IT.

Spatial staging of the work

If there are no alternative areas for IT, there will be an unwelcome clash between IT operations and the conversion work. It is strongly recommended to keep the two work areas separate as long and as well as possible. The focus here is on (spatial) staging: a specific area of the data center is cleared, separated with partition walls and then refurbished. On the one hand, this makes it easier to control the movement of people, and on the other, it minimizes the direct impact of the construction work on the IT system. Furthermore, the dust load for the data center can be reduced by means of negative or positive pressure.

Open heart surgery

Work within a productive environment is particularly challenging: If there is drilling dust, it is sucked in by the servers. If it drips, it can cause a short circuit. Vibrations are transmitted directly to the IT components. The "elementary" influences can still be (partially) measured and classified. In contrast, there are other sources of danger that are more difficult to classify: Restrictions in availability, risk of incorrect manipulation, non-IT personnel, sabotage/damage etc.

Maximum attention must be paid to work on the vital building services. If new cooling circuits are installed and the main electrical distribution system is replaced during operation, even minor faults can lead directly to a system failure. In this case, it is advisable to set up appropriate provisional solutions in advance. In this way, temporary redundancies can be created.

Conclusion

Conversions during ongoing data center operation are always a major challenge. Careful planning not only includes the structural areas, but also covers the following points in particular:

• Risk planning
• Defining levels of relapse
• Emergency planning (responsibilities, resources, etc.)

Particularly delicate work should be scheduled for the predefined maintenance windows. In general, a planned interruption is better than an outage during daytime operation. In terms of organization, it must be ensured that only authorized persons work in the security areas. Prior registration, compulsory identification and personal security checks are tried and tested means of achieving this. It must be specified which work is only permitted after prior notification and when accompanied. Close project support by the data center operator ties up resources, but is a basic prerequisite for constructive cooperation with planners and contractors. This is the only way to ensure successful conversion during "ongoing operation".