Delay measurements are important for network monitoring. With Internet monitoring platforms providing an unparalleled amount of data, it becomes necessary to automate their treatment. In particular, segmenting these delays permits supervising infrastructures and analyzing possible incidents. This paper explores the use of hierarchical clustering for the segmentation of multivariate network delays. The proposed method offers a computationally efficient way to identify the spatial correlation of network delay and to jointly segment time series within the same cluster. A post-treatment step is introduced that involves the Viterbi algorithm to smooth segmentation and handle the temporal dependency more effectively. This global method is evaluated on two real-world datasets, demonstrating its suitability for managing delays with varying variance and changing patterns. The proposed approach provides an efficient and cost-effective method for automated delay characterization.