The first detections of gravitational waves have opened an exciting new field of astronomy. One of the most fundamental limitations for the sensitivity of current and future interferometric gravitational-wave detectors is imposed by the quantum nature of light: Quantum vacuum fluctuations entering the interferometer through the readout port will contribute to the detection noise, at high frequencies in the form of shot noise and at low frequencies by radiation pressure noise. A promising way to reduce this quantum noise is the injection of squeezed states of light that have a lower uncertainty in one quadrature than the vacuum state. The GEO 600 gravitational-wave detector demonstrated the use of squeezed light in 2010 and it is now the first detector to routinely apply squeezing to improve its sensitivity beyond the limits set by classical quantum shot noise. This thesis details the practical aspects of long-term stable and efficient squeezed-light integration in a large-scale gravitational-wave detector. Imperfections that can limit the amount of observable non-classical noise improvement, such as optical losses and phase fluctuations, were studied in detail and methods for their mitigation were developed. Novel control schemes for the active stabilisation of the squeezed light field's phase and alignment were one main focus of the investigations. At the same time, important experience was gathered in the operation of the squeezed light source over long timescales. Over the course of the thesis work, improvements were implemented that significantly increased the performance of the squeezed-light application. Squeezing was injected with an overall duty cycle of 88%, reaching a noise reduction of up to 4.4 dB, corresponding to a 40% lowered shot-noise level. This work has firmly established the practical application of squeezing as a mature technology. The gained knowledge will directly inform the implementation of squeezed light for all future gravitational-wave detectors.