hpca » sample-final »

**Problem 5. [10 points] Disks and RAID**

A 6,000RPM disk drive has 5 platters, with 100,000 tracks per surface, 1,000 sectors per tracks, and 512 data bytes per sector. Each sector also has a 128-bit error detection code that can detect all errors in its sector but cannot correct any errors. The head can takes one microsecond (one millionth of a second) to move from one cylinder to an adjacent cylinder, and multi-cylinder movements are done at the same speed (one cylinder per microsecond). The disk controller is very fast (assume zero latency for everything it does) and the I/O bus has very large (assume infinite) bandwidth.

(A) [1 point] How many heads does the disk drive have?

(B) [2 points] Assuming that the disk controller and the drive itself are not servicing any other requests, what is the worst-case time needed to read a sector from the disk?

(C) [2 points] Assuming that the disk controller and the drive itself are not servicing any other requests, what is the best-case time needed to read the entire disk?

(D) [1 points] If we use two of these disk drives in a RAID0 configuration, what is the total data capacity of the resulting RAID array?

(E) [1 points] If we use two of these disk drives in a RAID0 configuration, can we recover all data if one sector is damaged on one of the disk drives? Explain your answer.

(F) [1 points] If we use two of these disk drives in a RAID1 configuration, can we recover all data if one disk drive is accidentally dropped into an active volcano? Explain your answer.

(F) [1 points] If we use five of these disk drives in a RAID5 configuration, what is the total data capacity of the resulting RAID array?

(G) [1 points] If we use five of these disk drives in a RAID5 configuration, is it possible to damage only two sectors in a way that the RAID array cannot recover from? Explain your answer.