Making global indexes UNUSABLE before a direct-path INSERT on a partitioned table prevents expensive index maintenance during the load, which is a major performance drain. Rebuilding them afterward is efficient. Dropping (Option A) loses the index definition and is more administrative overhead. Conventional path INSERT (Option C) would be slower for the load itself. SKIP_UNUSABLE_INDEXES (Option D) allows queries to proceed but doesn’t solve the slow load performance issue.

Oracle Database Vault uses Factors (like identity, time, IP) and Rule Sets to create complex conditions. Creating a ‘WORK_HOURS’ Factor that evaluates based on SYSDATE and including it in a Rule Set attached to a Command Rule is the standard, declarative method for time-based access control. A Realm Authorization (Option B) controls who can access objects in a realm, not necessarily when. A custom Command Rule alone (Option A) lacks the built-in logic for time. A Secure Application Role (Option D) is a database security feature, not a native Database Vault method for this specific time-bound policy.

In a LIST-partitioned table, if a row’s partition key value doesn’t match any explicitly defined partition list, it will be rejected unless a DEFAULT partition exists. The error indicates the value is not in any defined list, so the simplest fix is to add a DEFAULT partition to catch all unspecified values. Partition fullness (Option A) causes space errors, not ORA-14400. An UNUSABLE partition (Option C) causes errors on access, not on insert of new data. MAXVALUE (Option D) is relevant for RANGE partitioning, not LIST.

ALTER TABLE MOVE PARTITION is the direct and correct method to relocate a partition’s data to a new tablespace. Following this with ALTER TABLE … MODIFY PARTITION … READ ONLY changes the partition’s attribute. This operation maintains the partition’s logical structure and does not invalidate indexes if they are updated automatically (or can be rebuilt). Modifying the partition attribute (Option B) doesn’t move data. Online redefinition (Option C) is overkill for moving partitions. Exchange (Option D) is complex for this task and changes the object’s identity, which could break dependencies.

The ENCRYPTION_PASSWORD parameter is crucial. It creates a password-protected encryption key within the dump file set itself. During import on the target system, providing this password allows Data Pump to decrypt and re-encrypt the data with the target database’s wallet. Without this, encrypted data cannot be restored if the source wallet is not available. ENCRYPTION=ALL (Option D) encrypts the dump file, but doesn’t solve the key portability issue. INCLUDE=ENCRYPTION is not a valid clause. FULL=YES (Option C) specifies the scope, not the encryption method.

The -preferred/-available model directs all work primarily to the preferred node. Using the -uniform flag distributes connections evenly across all designated instances from the start, which is the appropriate fix for load imbalance under normal conditions. TAF (Option A) manages failover, not load distribution. Adjusting shared servers (Option C) addresses process limits, not the routing decision. Client-side CLB (Option D) helps, but the server-side service definition (-uniform) is the primary and most effective control for initial connection placement in RAC.

V$RESULT_CACHE_STATISTICS provides key metrics like ‘Create Count Success’, ‘Find Count’, and ‘Invalidation Count’. A high ratio of ‘Create Count Success’ to ‘Find Count’ or a high number of ‘Block Count Requires’ suggests that useful results are being aged out due to memory pressure, indicating a benefit from increasing the result cache size. V$RESULT_CACHE_MEMORY (Option D) shows current allocation, not advisory info. V$SGA_RESIZE_OPS (Option B) and V$MEMORY_RESIZE_OPS (Option C) show historical resize operations, not predictive benefits.

Flashback Data Archive (FDA) automatically tracks all historical changes to a table in a separate archive. It allows you to query the table AS OF any time within the retention period, making it perfect for precise, row-level recovery from human error without restoring backups. Flashback Query (Option D) works only if UNDO data is still available, which may not be long enough. RMAN PITR (Option B) recovers the entire tablespace and is disruptive. Flashback Table to BEFORE DROP (Option C) recovers a dropped table, not deleted rows.

For a balance of performance, storage efficiency, and availability, NORMAL redundancy (2-way mirroring) is typical for data. The key is to define failure groups that align with physical fault boundaries (like separate racks, arrays, or controllers). This ensures each mirror copy is stored on a physically independent device, providing true high availability. EXTERNAL (Option A) offloads protection but loses ASM’s awareness of storage topology. HIGH (Option C) provides more protection but at a significant storage cost, often overkill. Not defining failure groups (Option D) risks ASM placing both mirrors on devices that share a single point of failure.

Making global indexes UNUSABLE before a direct-path INSERT on a partitioned table prevents expensive index maintenance during the load, which is a major performance drain. Rebuilding them afterward is efficient. Dropping (Option A) loses the index definition and is more administrative overhead. Conventional path INSERT (Option C) would be slower for the load itself. SKIP_UNUSABLE_INDEXES (Option D) allows queries to proceed but doesn’t solve the slow load performance issue.0