Computing clusters at universities are often shared between many different user groups, having their own requirements concerning computing infrastructure, e.g. the operating system (OS), software packages, access policies and different grid middlewares. The integration of local computing and storage resources into the grid of a specific user group is realised by portal machines, providing different services. In the current implementation of the Worldwide LHC Grid and the CDF grid middleware, these services require dedicated operating systems and should be installed on different machines for stability reasons. Virtualisation of these services on one high-capacity machine eases maintenance of the systems and reduces the hardware overhead especially for smaller clusters. This approach of hardware consolidation is not restricted to the grid infrastructure and may therefore be applied to other even simpler services. This paper describes the process of hardware consolidation at a typical university grid centre. A discussion of different virtualisation techniques is also included and benchmarks show that Xen is an optimal candidate.