Abstract
The physical geometry of a modern magnetic disk drive does not correspond to its logical geometry. Unfortunately, existing disk simulators still use logical geometry of drives for simulation of scheduling algorithms. This results into inaccurate and unrealistic results. Therefore, this paper proposes physical-geometry-aware disk simulator to perform accurate and realistic simulation of scheduling algorithms. We conducted experiments to demonstrate the inaccuracy of simulation results for logical geometry using block trace of I/O workload generated by
Postmark
benchmark. The results suggest that in physical geometry of a drive, the cylinders change more frequently and the seek length covered with each cylinder change is significantly more as compared to logical geometry. Moreover, though cylinder accesses in both geometries stay localised within a range of cylinders, this range is wider in physical geometry as compared to logical geometry. Furthermore, FCFS, SSTF and SCAN incur 55%, 44% and 122% more seeks, respectively, when simulated using physical geometry as compared to logical geometry. This implies that the simulation of scheduling algorithms using logical geometry is neither accurate nor realistic. We also demonstrate the usefulness of the tool in academics using the Technology Acceptance Model (TAM). The results suggest that the participants (
n
=
83
) showed high-positive intention to use the tool (
I
M
=
5.39
, SD = 0.91,
α
=
.
80
), high-positive attitude towards the tool (
A
M
=
5.46
, SD = 0.79,
α
=
.
69
), high-positive perception of the usefulness of the tool (
P
u
M
=
4.05
, SD = 0.74,
α
=
.
70
) and high-positive perceived ease of use of the tool (
P
e
M
=
3.97
, SD = 0.61,
α
=
.
77
).