Introduction to CMS from a CS viewpoint

Most of this text follows the ‘main stream’ of CMS very closely. However I have not always marked places where the document goes beyond ‘main stream’ CMS thinking. The thesis deals with the development of an efficient, self-optimising storage layer for event data. The layer is intended to manage storage space located in a single computer centre, so wide-area issues are not discussed in depth below. The issue of having a unique identity for each stored object is addressed. Taking this as a starting point, a generalisation to unique identities for virtual data objects is straightforward, however the text below does not address such a generalisation.