32 Caching

Disk blocks used for storing directories or recently used files caches in main memory. Block periodically written to disk
$\Rightarrow$ Big efficiency gain
Inconsistency arises when system crashes.
Reason why computers must be shutdown properly

Journaling File Systems

To minimise data loss at system crashes, ideas from database are used :

• Define Transaction points : Points where cache is written to disk.
  $\Rightarrow$ Have consistent state
• Keep log-file for each write operation Log enough information to unravel any changes done after latest transaction point

Disk Access

Disk access contains three parts :

• seek : head moves to appropriate track
• latency : correct block is under head
• transfer : data transfer
  HDDs : Time necessary for seek and latency dwarfs transfer time
  $\Rightarrow$ Distribution of data and scheduling algorithms have vital impact on performance for HDDs, less so for SSDs