IS215 - File and Disk Management Exercises - 2

11. What method of ensuring file systems reliability is suitable for each of the following 'file system:

a) small in size and data loss is tolerable (quite cheap to recover)
b) small and data loss are very expensive, hence it should be avoided
c) large in size and data loss is tolerable
d) large in size and data loss is very expensive

Explain your answer.

12. Periodic dump is expensive. Explain.

13. Discuss the procedure for recovering from file system crash, when the system used is a combination of periodic dump, incremental dump and transaction logging.

14. Name an actual system that uses the directory structure:

a) single-level directory
b) two-level directory structure
c) tree-structured directory
d) acyclic directory

15. Discuss a method of checking the consistency of a file system.

16. During file system recovery after a crash, files are often compacted into contiguous regions on the disk even though such contiguity may be irrelevant to the file access method. Explain why compaction may still be advantageous.

17. Given the following requests for disk access in a disk whose tract number ranges from 0 to 99.

20, 7, 99, 56, 75, 19, 89, 2, 50

The read/write head is currently reading track number 15 and has just gone from track number 26, if the scheme used is:

a) First-Come-First-Serve
b) Shortest-Seek-Time-First
c) SCAN
d) CSCAN
e) N-Step SCAN

18. Differentiate non-contiguous sector allocation from non-contiguous block allocation.

19. What are the overheads involved with the allocation strategies: contiguous, linked, 'and indexed, if the file:

a) grows at the beginning
b) grows at the middle
c) grows at the end?

Answer also the cases when the file shrinks instead of grows.

20. All the disk scheduling algorithms (SSTF, SCAN, C-SCAN, N-Step SCAN) except FCFS are not truly fair (starvation may occur). Arrange the algorithms in order of degrading fairness. Show how you arrived at your answer.