Key figure controlled data center

Data & Analytics

Key figure controlled data center

For successful strategic corporate management, precise knowledge of the facts in one's own corporate processes is just as important as statistically processed knowledge of the concrete degree of realization of qualitative corporate goals. noventum consulting examines how key figure systems can support the management of data centers and IT services.

Systems and methods for corporate management

For many years, ITIL (IT Infrastructure Library) has been one of the well-known frameworks with a mature control method. ITIL stands equally for processes and procedures and consistently maintains the perspective of IT service management. In ITIL itself, key figures are noted for all processes. These are pure process key figure series, which are not directly assigned to targets. The key performance indicators from the ITIL documentation cannot fully answer the question of whether IT and business objectives are supported.

The COBIT framework (Control Objectives for Information and Related Technology) divides the tasks of IT into processes and control objectives and provides control specifications that primarily define what is to be implemented, but not how this is to be done. It thus primarily satisfies the requirements of IT governance, which demands that statements be made about the quality of goal achievement. To this end, COBIT provides a very useful and good structure and is recognized for its holistic mapping of workflows and processes. COBIT is therefore the right choice as a basis for controlling according to key performance indicators.

Target hierarchy

This can be seen particularly clearly in the goal hierarchy presented by COBIT. Here, COBIT defines that IT is subordinate to business goals and that it is there precisely to support the business - a statement that is also included in ITIL. In addition to this statement, COBIT provides a hierarchy of 17 Business Goals, 28 IT Goals and 34 Process Goals, which are underpinned by activities and key performance indicators. This hierarchy can be used both in parts directly and as a starting point for your own goal definitions.

However, when further elaborating the COBIT specifications, one suffers a slight shock - COBIT defines 340 different metrics for the goal hierarchy. Now the question arises whether a data center can and should be managed with 340 metrics. A pilot of a Boeing 747-300 has to cope with 971 instruments and displays. In more modern aircraft, this flood is reduced to about 300, which allow access to even more information by means of flight management systems. Reduction of complexity is also required in our topic.

For the management of a data center, it is therefore necessary to limit oneself to the key figures that are meaningful at the moment and for the company, and which can also be determined with a reasonable amount of effort. Even after a reduction to what is sensible and feasible, the list of key figures is still long. Therefore, a mapping to target achievement and the presentation of the data in a cockpit makes sense.

Comparing apples with oranges

In order to aggregate the metrics into a meaningful statement of target achievement, there are some difficulties to be solved. How is a metric evaluated, which values are "good" and which are "bad". How are the various metrics and sub-goals aggregated? Is a weighted average better, or should it be a minimal view? How do you compare metrics that are at different temporal granularities?

In addition, there is the desire to evaluate the contribution of a business unit, customer, or service to the achievement of goals. This means that for as many of these metrics as possible, these dimensions must be included and taken into account when setting thresholds.

Data Center Cockpit

In order to test the implementation of these control principles, noventum consulting GmbH started a master project together with the DBIS group of the Westphalian Wilhelms University in Münster (see also the interview with Prof. Dr. Vossen on the following pages). The task for the students was to develop the prototype of a management cockpit, which transfers the hard facts of the key figures into the soft target definitions by means of calculation rules. The Data Center Cockpit should also be usable for different use cases, so that, for example, a service manager can look at his cut of the data as well as a key accountant or service level manager.

Further procedure

noventum consulting will continue to work on the KPI-driven management of data centers and IT departments. The next steps include an extension of the current platform, a project to optimize the possible metrics and a realistic field test.

Selected customers have already expressed interest in the Data Center Cockpit, so we are confident in expanding not only the topic, but also the toolset.

Sorry, item "1187" does not exist.


Marc Buzina

Go back