NCM-DB study plan

Distinguish the instance from the database and explain how memory structures and background processes cooperate to serve a session.

Compare logical storage hierarchy (database, tablespace, segment, extent, block) with physical storage (data files, redo logs, control files) and describe how they map to each other.

Explain the read-consistent behavior of a query against a concurrently modified table, including undo and redo roles in that scenario.

Draw and label the startup sequence from instance start through database open, noting which structures are read at each stage.

Design a small normalized schema for a plausible business case, then deliberately denormalize one part and justify the trade-off.

Choose primary keys, foreign keys, and indexes for that schema and state the selectivity reasoning behind each index decision.

Describe how a transaction boundary and isolation level choice would change behavior for two concurrent writers on the same rows.

Identify which integrity constraints belong in the database versus application code for the same schema.

Install the database engine on a supported platform and verify the instance starts cleanly from a cold boot.

Configure a second instance with a non-default port, memory allocation, and character set, then confirm both run simultaneously.

Create a tablespace with a specified data file size and autoextend policy, then create a user whose default tablespace is that new tablespace.

Explain what each of the three core parameter categories (memory, file location, process/session limit) changes at instance start versus at runtime.

Create and alter the main schema object types: table, index, view, sequence, and a constraint, then verify each in the data dictionary.

Demonstrate what happens to dependent objects when a parent table is dropped, and how to reclaim space afterward.

Compare heap-organized and index-organized storage for a lookup-heavy table and state when each is preferable.

Explain extent allocation and space reclamation behavior when rows are deleted but the table remains in use.

Write multi-table queries using inner and outer joins, aggregation with GROUP BY and HAVING, and a subquery in both the SELECT list and the WHERE clause.

Use window functions to rank rows within groups and compute running totals without collapsing the result set.

Read an execution plan and identify the join order, access method per table, and the most expensive step.

Explain how the optimizer's statistics influence plan choice and what happens to plans when statistics are stale.

Take a consistent full backup and a separate archive log backup, then restore both into a new location and verify row counts match the source.

Perform a point-in-time recovery that intentionally discards the last few transactions, and explain which log sequence determined the stop point.

Describe the difference between a cold backup and a hot backup and the recovery preconditions each imposes.

Outline the failover path for a standby configuration, naming the steps and the failure modes that would leave the standby unusable.

Capture a baseline of the top wait events and the top SQL by elapsed time during a synthetic load, then identify the single largest contributor.

Resolve one diagnosed performance problem, such as a missing index or an undersized memory area, and re-measure to show the change.

Create roles and users with least privilege, grant object privileges through roles rather than directly, and verify effective privileges from the user's session.

Enable auditing for a specific privileged action and confirm the audit trail records both a successful and a failed attempt.

Re-run a timed 60-question mixed practice set spanning all seven domains and score it by domain.

Re-teach the two weakest domains from your own notes without opening the study material, then check against source notes.

Review every missed question and write the underlying concept in one sentence each, not just the correct answer.

Confirm you can state the architecture, backup, and tuning fundamentals aloud in under ten minutes total.