Data center: first and foremost a production site
Unlike manufacturers’ plants, data centers are often operated piecemeal, with a lack of documented methodologies and procedures. Given the increasing criticality of information systems, isn’t it time for firms to industrialise the operation of their technical infrastructures?
Data centers: an imbalance between design and operation
Most firms devote considerable resources to building new data centers or refurbishing property assets to house them. Day-to-day, however, they ensure they can deliver the level of service their users expect by operating their buildings and the hardware they contain effectively. So, over the lifetime of a data center (10 to 20 years), it will cost more to operate than design.
Nonetheless, the “operations” element is too often the poor relation of design specifications, because in this area cost control still largely takes precedence over risk control.
Operations: the key ingredient in a data center’s continuity of service
In practical terms, there are two dimensions to data center operations: on the one hand, the management of regular upgrades of IT infrastructure – both physical and logical – by IT teams. And on the other hand, management of the building and its fluid technical infrastructure (power supply, cooling, etc.) by facilities management or the property department.
In the manufacturing sector, this dual management role has been performed for many years: the everyday operation of factories involves documented installation, maintenance and incident resolution procedures. When it comes to data centers, it is still quite rare. However, operating data centers using an industrial approach produces the same effects: better overall steering of investments, standardisation of working methods and operational processes (maintenance, daily management, security, protocols in the event of malfunction, etc.) and improvement of the quality of service.
Industrial data center operations: a profitable investment
The industrialisation of data center operations requires investment – of time but also money. It requires good coordination between the teams responsible for IT and for managing a company’s property assets. In total, it generally takes around one year to draw up exhaustive specifications, encompassing an inventory of facilities and the definition and finalisation of procedures, including testing different scenarios.
This investment needs to be put into perspective with the benefits for a company: the implementation of standardised, reliable procedures helps limit the number of incidents, in particular incidents associated with human error. Documented protocols also go hand in hand with KPIs tracking compliance with service levels and maintenance response times, capacity increases, the restoration of service following an incident, etc.
These data can be compiled in dashboards, enabling better overall monitoring of all the data centers belonging to a company or group. They also make it easier to manage the relationship with the provider responsible for multi-technical maintenance (PMT), when the operations role is contracted out: the procedures rely on precise protocols, defined before the contract is signed, and the service level agreements can be monitored on the basis of detailed KPIs.