Database Management Systems

A snowflake schema extends the star schema by normalizing dimension tables into multiple levels. This introduces all three table types: fact tables (central), dimension tables (primary and secondary), and helper/bridge tables (additional normalization levels). The 'helper' tables are the normalized-out dimension attributes.

The 'invalid object or does not exist' error in the APPS schema occurs when the database synonyms or grants pointing from the APPS schema to the underlying product schemas (like GL) are missing or broken. Recreating grants and synonyms using the adadmin utility restores these links. Relinking, patching, or reinstalling is premature.

In Oracle Applications 11i, the Oracle 8.0.6 database files and iAS (Oracle Internet Application Server) files are stored in the ORA directory. The DB directory doesn't exist in the standard structure. COMN contains common files, FND_TOP is for Application Object Library, and APPL_TOP contains product-specific application files.

When comparing datasets with utilities like IEBCOMPR, both datasets must have the same record length (LRECL) for a valid comparison. The block size (BLKSIZE) can differ, but the logical record structure must match for byte-by-byte comparison.

Cataloged datasets have their volume serial numbers (VOL=SER) recorded in the system catalog, so the operating system can locate them automatically. Uncataloged tape datasets must specify the volume serial number explicitly so the system knows which tape to mount.

DB2 can be accessed from all these environments: TSO (Time Sharing Option) for interactive use, CICS (Customer Information Control System) for transaction processing, IMS (Information Management System) for database management, and BATCH for scheduled jobs. Each is a valid DB2 access method.

Normalization is the systematic process of organizing fields and tables of a database to minimize data redundancy and dependency by dividing large tables into smaller, linked tables.

A data dictionary is a comprehensive repository of metadata about the database structure, maintained by the database server itself. It contains information about tables, users, privileges, constraints, and all database objects. Option E is correct because the data dictionary is indeed a collection of tables, contains database information, and user information.

Fast Load has specific constraints: it permits at most one data type conversion per column for performance, requires all AMPs to be online for restartability, and loads exactly one empty target table per job. It does not support triggers during load operations.

Fast Load restartability requires maintaining the state across runs. The target and error tables must exist before the script runs and persist after. Dropping them would lose restart information. Checkpoints provide recovery points so FL can resume from the last successful point rather than starting over.

Tpump offers flexible real-time loading: configurable update rates (A), continuous background operation with flow control (B), support for up to 60 target tables from one source (C), safe stoppability without lock issues (D), and efficient SQL execution via macros (E). It is not designed for the extensive error handling that Multi Load provides.

When the Oracle database engine automatically opens and manages a SQL cursor for its internal processing of DML statements or queries, it is known as an implicit cursor. Explicit cursors, conversely, are declared and named by developers in PL/SQL.

Source Analyzer in Informatica can import or create multiple source types: relational tables/views/synonyms (a), COBOL files (c), XML files and data models (d). However, fixed-width and delimited flat files containing binary data (b) are NOT supported through Source Analyzer - binary data requires different handling. Therefore the correct combination is a, c, and d.

Tracing levels control logging granularity. Normal provides basic info, Terse gives minimal logging, Verbose Initialization logs startup details. Verbose Data writes a log entry for EVERY single row processed, which creates maximum overhead and significantly slows data loading performance. This level is typically used only for debugging specific row-level issues.