💠 6) I/O & stockage (ASM, latence, redo, temp, cache)
Objectif : relier les symptômes “DB lente” à la réalité I/O : latence read/write, comportement redo/commit, TEMP spill, cache hit/miss, et architecture stockage (ASM, diskgroups, striping, path).
Latence I/O : lire les “vrais” signaux
db file sequential/scattered read, read latency vs throughput, files vs system.
Redo / Commit : log file sync & friends
LGWR, commit latency, log buffer, log file parallel write, storage redo path.
TEMP : sorts, hash joins & spills
temp usage, workarea, PGA insuffisante, spill-to-temp, symptômes et remèdes.
Cache : buffer cache & smart usage
Logical vs physical reads, cache hit ratio (avec prudence), hot blocks, tuning mindset.
ASM & diskgroups : architecture & pièges
Striping/mirroring, allocation units, rebalance, I/O distribution, monitoring.
Runbook I/O : diagnostiquer “storage vs SQL”
Méthode standard : preuve, isolation, mapping files, tests, actions, rollback.
Diagrammes (Mermaid-ready)
Chaîne I/O → waits → symptômes, et décision redo/temp/cache.
Commandes & vues (I/O)
V$SYSTEM_EVENT, V$FILESTAT, V$IOSTAT_*, ASM views, TEMP usage, redo stats.
