Filesystem yana mai da array mai lebur na disk blocks zuwa matsayi na files da directories, yana sarrafa metadata da sarari na kyauta don sa hakan ya zama abin dogaro kuma mai sauri.
A filesystems na UNIX, kowane file ne (index node) — tsari da ke riƙe na file (girma, mai shi, izini, timestamps, link count) da . Lura da abin da inode adanawa: . Sunaye suna zama a cikin , waɗanda kawai files ne masu danganta suna → lambar inode:
directory "docs" inode 1042 data blocks
┌───────────────────┐ ┌──────────────┐ ┌────────┐
│ "report.txt" →1042│───────────▶│ mode, uid │──ptr───▶│ block │
│ "notes.txt" →1050│ │ size, times │──ptr───▶│ block │
└───────────────────┘ │ link count=2 │ └────────┘
│ block ptrs │──(indirect)──▶ ...
└──────────────┘
Shi ya sa hard link kawai entry na biyu na directory ne mai nunawa zuwa inode iri ɗaya (link count 2), kuma dalilin da ya sa rm da gaske unlink ne — yana cire suna kuma yana 'yantar da inode ne kawai idan link count ya kai sifili.
Aiki guda na hankali (misali append) yana taɓa blocks da dama — inode, free-space bitmap, da data. Crash a tsakiya yana barin filesystem cikin rashin daidaito. Journal da farko yana rubuta canje-canjen da ake nufi zuwa log, sa'an nan yana amfani da su; bayan crash FS yana sake kunna (ko watsar) journal maimakon cikakken scan mai jinkiri (fsck). Modes suna cinikin aminci da sauri: journal na metadata kaɗai (mai sauri, na kowa) da metadata da data.
Kernel yana cache ɗin bayanan file a RAM a cikin page cache, don haka sake-karatu suna buga memory kuma rubuce-rubuce ana buffered ana kuma flush ɗinsu daga baya (write-back):
buffered I/O (default): app ↔ page cache ↔ disk → cached, async writeback, fast re-reads
direct I/O (O_DIRECT): app ↔ disk → bypass the cache; DBs manage their own
fsync(): force buffered data to durable storage (required for real durability)
Buffered I/O yana da sauri, amma bayanan da aka "rubuta" ba su da dorewa har sai fsync; databases sau da yawa suna amfani da direct I/O tare da nasu buffer pool don sarrafa caching da dorewa daidai.
Cikin filesystem yana bayyana halayen yau da kullum: dalilin da ya sa hard links ke aiki, dalilin da ya sa share file da process ke ci gaba da riƙe a buɗe ba ya 'yantar da sarari sai ya rufe, dalilin da ya sa fsync ke da muhimmanci ga dorewa (kuma bug na gargajiya na asarar bayanai idan aka bar shi), da dalilin da ya sa karatu na biyu na file yake nan take (page cache). Injiniyoyin senior suna tunani game da waɗannan sa'ad da suke zaɓar buffered vs direct I/O ga database, daidaita girman RAM don cache, ko debug ɗin matsalar crash-consistency.
Ɗakin karatu na tambayoyin hira na IT tare da amsoshi cikakke — daga Junior zuwa Senior.
Ba da Gudummawa