The two databases are independent, and their interface is limited: just the data structure that contains the keys is in common between the two.

This improves the modularity of the design. As a consequence, for instance:

• the part that describes the database structure is independent from observation schemas (which are subject to refinements);
• the design of the two parts can be optimized for the specific operations (replicas, RO vs. RW, distribution);
• the technologies used for the two parts are independent: one can switch one from SQL to WML or LDAP without touching the other.

We envision two kinds of scenarios, and have designed the Perl APIs to enable a simplified access in Perl.

next: Scenario 1

previous: Work done

up: JTB phone call support (12 May 2003)

index