SQL/MX 3.2.1 Management Manual (H06.26+, J06.15+)

Versioning Capabilities
The nodes in a distributed SQL/MX database can run different releases of SQL/MX software. Each
release of SQL/MX software can identify the other releases that it is compatible and incompatible
with. The SQL/MX versioning design ensures that:
• Nodes with compatible software versions can interoperate.
• Access to or from nodes with incompatible software versions fail predictably.
SQL/MX Release 3.x does not support interoperability with SQL/MX Release 2.x in a distributed
database environment. You cannot perform queries, through embedded SQL or dynamic SQL,
from a node running SQL/MX Release 3.x on a node running SQL/MX Release 2.x, and vice
versa.
For more information, see the SQL/MX Release Installation and Upgrade Guide.
User Data Distribution
You can distribute tables and indexes across nodes in an Expand network or allow them to reside
intact on single nodes.
User database objects, including base tables, indexes, and views, are distributed with the
LOCATION clause in DDL and utility operations. Distribution of user data is an explicit user action.
NonStop SQL/MX does not impose limits on the number of nodes to which user database objects
can be distributed.
Catalog Reference
SQL/MX database objects have location-independent ANSI names. The actual data is represented
in Guardian files with location-dependent Guardian file names. The information needed to resolve
an ANSI name to the corresponding set of Guardian file names (ANSI name lookup) is stored in
SQL/MX metadata.
Metadata definitions for database objects are stored in metadata tables in definition schemas.
One definition schema holds the metadata for all schemas of a given version in a catalog. A DDL
or utility operation can add, remove, or modify rows in the appropriate definition schema for the
affected database object's catalog.
You use catalog references to identify the node on which the metadata for a given catalog resides.
Several types of catalog references are used. An automatic catalog reference for a catalog occurs
on the node where the catalog is created and the metadata for all the database objects in that
catalog resides. A given catalog can have only one automatic reference node in the distributed
database. An empty (or manual) catalog reference occurs for every node on which the catalog
has been registered from node where the catalog was created. The system schema tables on empty
catalog reference nodes contain information about the catalog and its node that enables these
nodes to locate and query objects in the catalog. A node on which a catalog has not been created
or registered contains no catalog reference cannot detect the catalog, its objects, or its metadata.
SQL/MX Release 3.x supports catalog reference but not object metadata replication. A catalog
reference can be registered for remote nodes, providing them with the catalog registration metadata
needed to access the catalog’s object metadata on the node where the catalog was created.
Object metadata for any user catalog, including all metadata definitions, remains on the node
where the catalog was created and cannot be replicated to other nodes in the network.
ANSI Name Lookup
ANSI name lookup encompasses:
• ANSI-name-to-Guardian name resolution
• Metadata definition lookup using ANSI names
When you register a catalog from the local node to remote nodes, you ensure that the catalog’s
database objects can be successfully resolved to corresponding underlying Guardian file names
268 Managing an SQL/MX Distributed Database