Clustering Using A DB Which Already Contains The Distances Between Nodes

Hi,

I've got quite a simple case study, I've got a database which contains N nodes and a table containing N*(N-1) rows with the distances between each two nodes (it's a symmetrical matrix), I need to do several iterations and say to which clusters and sub-cluster each of the nodes belongs to, I failed to understand if it's within the SQL SERVER capabilities to handle already known distances in order to perform the clustering algorithm (the last part of it, just combining the groups according to the principle of minimum distances within each cluster and maximum distances between the different clusters).

In addition I would also need to know the followings:


is it possible to create an iterative process which will create several levels of clustering (number of groups found), and if so how do I tell the Mining tools how many group to create?

is there an over-lapping clusters process in the SSAS? (I.E if one node belongs to several clusters who are not a complete sub-groups of each other)

Thanks!
Ron