1 Introduction

The Centre de Données astronomiques de Strasbourg (CDS) has a very long experience in acquiring, cross-identifying, and distributing astronomical data (Genova et al. [2000]): a collaboration for the exchange of what was called machine-readable astronomical data started with the NASA-GSFC and the Astronomisches Rechen-Institut around 1970. This collaboration has been maintained over this 30 year period, and collaborations with other institutes for similar exchanges have been developed. The volume of data shared of course increased, at a rate which has been exploding in the recent years.

Compared to the late 60's, where the bulk of the machine-readable data consisted in a set of the basic catalogues carefully keypunched, the situation has changed drastically, now that every instrument or detector is generating megabytes or gigabytes of daily output. These huge data sets are hopefully not stored in data centers, but are processed in the observing center where the expertise exists to generate the best high-quality archives and catalogues in a form usable by astronomers who are not familiar with the instrument. The Data Centers' role is essentially to collect such "final'' catalogues, or more generally high-quality data, i.e.data which either were published in the refereed scientific literature, or at least a paper describing these data and their context was accepted for publication in a refereed scientific journal.

Making an efficient usage of the data distributed by the data centers -- for instance for the analysis of the statistical properties of some interesting population of stars -- often requires to combine data coming from several data sets; this operation is far from simple, and this is why the first creation of CDS was SIMBAD, a data-base resulting from the cross-identification of the major catalogues, later expanded to thousands of catalogues and to published literature (see Wenger et al. [2000]).

The VizieR system results from a different approach: the astronomical catalogues are kept in their original form, but homogeneous descriptions of all these data sets are provided in order to maximize their usability. In other words, VizieR relies on an homogenization of the catalogue descriptions -- what is also called metadata, or data describing other data -- to transform the set of machine-readable astronomical catalogues into a set of machine-understandable data. VizieR actually consists in an interface able to query this set of machine-understandable astronomical catalogues.