Describe the extensions required to query processing and query optimization to fully support the ORDBMS.

What will be an ideal response?

Some possible answers are as follows:
One capability we require is that the ORDBMS query processor flattens queries whenever
possible. If the function has been defined as an external function it may be to provide an
extensible query optimization mechanism. This may require the user to provide a number of
routines specifically for use by the query optimizer in the definition of a new ADT.
Another strategy that should be possible is to allow a UDF to be invoked not from the
ORDBMS server, but instead from the client. This may be an appropriate strategy when the
amount of processing in the UDF is large, and the client has the power and the ability to
execute the UDF (in other words, the client is reasonably heavyweight). This alleviates the
processing from the server and helps improve the performance and throughput of the overall
system.
An ORDBMS can compute and index the result of a user-defined function that returned scalar
data (numeric and character data types). Traditional relational DBMSs use B-tree indexes to
speed access to scalar data. However, a B-tree is a one-dimensional access method that is inappropriate for multidimensional access, such as those encountered in geographic
information systems, telemetry, and imaging systems. With the ability to define complex data
types in an ORDBMS, specialized index structures are required for efficient access to data. A
mechanism to plug in any user-defined index structure provides the highest level of flexibility.
This requires the ORDBMS to publish an access method interface that allows users to provide
their own access methods appropriate to their particular needs. Although this sounds relatively
straightforward, the programmer for the access method has to take account of such DBMS
mechanisms as locking, recovery, and page management. An ORDBMS could provide a
generic template index structure that is sufficiently general to encompass most index.