I/O performance of the prod database on your Exadata Database Machine has degraded slightly over the past month. The database has been allocated to the OLTP I/O Resource Management (IORM) category…

Question

Options

• AOS I/O metrics using Enterprise Manager host pages for the storage servers
• BOS I/O metrics using OS tools such as iostat on the database servers
• CI/O-specific dynamic performance views such as v$iostat_fiie, v$iostat_function, and
• Dcellcli (or exacli/exadcli) to examine storage server metrics such as database, category, ceiidisk,
• EOS I/O metrics using OS tools such as iostat on the storage servers

• A
  13% (9)
• B
  21% (14)
• C
  59% (40)
• E
  7% (5)

Explanation

Options C and D are correct because they target the two layers where Exadata I/O problems actually live: the Oracle database layer and the Exadata storage cell layer. The v$iostat_file and v$iostat_function views (C) expose granular, database-scoped I/O statistics directly from Oracle's perspective, making it easy to see which files or functions are consuming I/O. cellcli/exacli (D) is the native Exadata storage cell CLI that can query metrics broken down by database, IORM category, and physical disk - exactly what you need when IORM is involved.

Options B and E (iostat on database servers and storage servers respectively) are wrong because iostat is a generic OS tool with no awareness of Oracle databases, IORM categories, or Exadata internals - it can't tell you why I/O degraded or which database/category caused it. Option A is similarly limited: EM host pages for storage servers show OS-level metrics, not the Exadata-specific or IORM-aware statistics you need for this investigation.

Memory tip: On Exadata, think in two layers - Oracle layer (v$ views) and Cell layer (cellcli). Generic OS tools like iostat sit outside both layers and are blind to IORM categories, making them the wrong choice whenever the question involves Exadata-specific or database-specific I/O analysis.

No community discussion yet for this question.