In a SQL scenario, the structure of the Grid is not bound to the hierarchical tree induced by an LDAP tree.
Therefore we need to introduce a structure, that will be used to index the database and improve performance.
The case of resources that are bound to a node is quite different from the case of resources that represent communication between nodes: the former grow with the number of nodes, the latter with its square.
As a consequence, representing communication resources using end-to-end characteristics may limit the scalability of a monitoring architecture.
In addition, availability of end-to-end mesurements implies that:
Therefore it would be appropriate to partition the nodes in the Grid so that the number of partitions is remarkably smaller than the number of nodes: we call domains these partitions.
The monitoring activity should be limited to communications between
domains.
next: GlueDomains
previous: Store observations in a SQL database
up: Seminar May 16, 2003