The new generation information technology (IT) services like mobile Internet, Internet of things (IoT), cloud computing, processing of big data, applications of artificial intelligence, etc. are becoming popular with the development of the information and communication technology (ICT) industry. In this industry, the dependency on the data centers is also increasing to ensure the quality of services (QoS). Thus, the energy consumption of the data centers is increasing with the increasing demand for computational resources in it because the load sections of the data center with sensitive equipment run $24$ hours a day, $365$ days of the year. Regarding data center operation, it is becoming a technical challenge to make a trade-off between reducing the energy consumption to limit the operational costs and ensuring higher reliability of the data center.

A way to help data center operators to cope with the posed challenges is by identifying the ``right size of the computational resource'', considering the power losses and service availability of the data center. This endeavor requires power consumption models that can consider different load sections with different types of equipment. The power consumption models of the load sections can address the electrical load demand and the power losses, especially losses in the internal power conditioning system (IPCS). On the other hand, the service availability of the data center mainly depends on the availability of the computational resources like servers and on the availability of the power supply through the IPCS. It is important to characterize the servers' failure and repair times to develop the stochastic model of the server unavailability in operation. The availability of adequate power supply through the IPCS depends on its component failures and the power supply capacity of its components. The bottleneck of the power supply capacity of the IPCS is subjected to the power losses of the equipment in the IPCS. Additionally, the voltage disturbances like voltage dips and swells in the IPCS also interrupt the power supply units (PSUs) of the servers, which also degrades the QoS of the data center.

The outcomes of this thesis can be synthesized as follows: 1) A comparative analysis of the energy consumption models of the major load sections in the data center, and an analysis of the impact of the power losses in the IPCS on the outage probability of the servers. 2) Reliability indices to assess the adequacy of the computational resources in the data center considering the outages of power supplies and the servers in operation. 3) The impacts of voltages disturbances in the IPCS on the power supply outages, hence on the interruptions of servers. 4) An analysis of the trade-off between the energy efficiency and reliability in operational planning of the data center.