SQL/MX 3.2 Management Manual (H06.25+, J06.14+)

Table 21 (page 273) describes the status of each node in the network with respect to user data,
object metadata, applications for CAT_1 and CAT_2, and the effects of the loss of each node on
the other two nodes in the network:
• The column “CAT_1 and CAT_2 User Data Present?” identifies whether user data for CAT_1
and CAT_2 objects is present on a given node in the network.
• The column “CAT_1 and CAT_2 Metadata Present?” identifies whether object metadata for
CAT_1 and CAT_2 is present on a given node. Only node \B, where the catalogs CAT_1
and CAT_2 were created, contains this object metadata because catalog and schema
replication is not supported with SQL/MX Release 3.x.
• The column “CAT_1 and CAT_2 Applications Present?” identifies whether application programs
that contain queries to CAT_1 and CAT_2 objects are present on a given node.
• The column “Consequences of Losing This Node on the Other Nodes” describes how the loss
of a given node affects the other two nodes in the network.
Table 21 Consequences of Network Node Loss, Original Configuration
Consequences of Losing This Node on the Other Nodes
CAT_1 and
CAT_2
CAT_1 and
CAT_2CAT_1 and
CAT_2 User
Data Present?Node
Applications
Present?
Metadata
Present?
From nodes \B and \C, you cannot perform DDL or utility
operations on objects partitioned across node \A. You cannot
YesNoYes\A
perform statically compiled DML operations that require data
stored on node \A.
From nodes \A and \C, you cannot perform DDL or utility
operations on CAT_1 or CAT_2. You cannot perform
YesYesYes\B
dynamically-compiled or automatically-recompiled DML operations
on any CAT_1 or CAT_2 objects. You cannot perform
statically-compiled DML operations that require data stored on
node \B or late name resolution of ANSI names.
From nodes \A and \B, you cannot perform DDL or utility
operations on objects partitioned across node \C. You cannot
YesNoYes\C
perform staticallycompiled DML operations that require data stored
on node \C.
Table 21 (page 273) shows that:
• User data for CAT_1 and CAT_2 objects resides on all three nodes in the network.
• The metadata for CAT_1 and CAT_2 resides only on node \B, where the catalogs were
created.
• Applications that contain queries to CAT_1 and CAT_2 objects reside on all three nodes.
• If the node that contains a catalog’s metadata (node \B for CAT_1 and CAT_2 metadata)
becomes unavailable, the surviving nodes cannot execute dynamic queries on the catalog
object data.
Note that the loss of the node where the object metadata resides does not prevent statically-compiled
queries from being performed in application processes on the surviving nodes to access local data.
However, they can query only data located on the local node or some other surviving node.
A minor reconfiguration of the distributed database environment can alter the impact of a node
outage on the other nodes. Table 22 (page 274) describes a modified configuration in which only
node \B contains application programs that query CAT_1 and CAT_2 objects as well as containing
the CAT_1 and CAT_2 metadata. In this revised configuration, node \B becomes more of a single
point of failure for the distributed database.
Managing a Network-Distributed SQL/MX Database 273