The scheduling or allocation of images to nodes of a containerized computing environment is provided. Images that share layers (i.e. images that have one or more layers in common) are allocated to the same node. To determine whether different images share one or more layers, leveraging metadata that is associated with the images is provided. By analyzing metadata associated with the images, it may be determined if the images have one or more layers in common (i.e. comprise the same layer(s)). In this way, information that is already available with conventional images may be used.