Diagnosis and Performance Monitoring - Deploying and Managing a Cloud Infrastructure - page 142

TABLE 5.2 Spindle rpm and associated latency

Spindle rpm Average Rotational Latency (ms)

4,200 7.14

5,400 5.56

7,200 4 .17

10,000 3.00

15,000 2.00

Data Transfer Rate

Data transfer rate is the effective throughput of the disk drive; it covers both the internal
and external data transfer rates. The internal transfer rate has to do with the speed at which
the drive moves data between the disk's surface and the controller of the drive. The external
data rate is the speed at which data is transferred between the drive's controller and the host
system. Since a system is only as fast as its slowest component, the data transfer rate will be
whichever is the slower of the two rates. The sustained throughput or sustained data transfer
rate will also be either the sustained internal or the sustained external data rate, whichever
is slower. The internal data transfer rate is determined by the sector overhead time, cylinder
switch time, head switch time, and media rate.
Sector Overhead Time This is the additional time the disk needs to process sector-related
data, which is needed for the control structure of the drive and includes information used
in drive management like locating and validating data in sectors and for various other sup-
port functions. This is similar to the processing of TCP/IP headers in a network frame.
Cylinder Switch Time This is the time needed to switch between data surfaces and the
read/write head because most typical HDDs have more than one disk and hence more
actuators and heads corresponding to each surface. This typically takes 2 or 3 ms.
Head Switch Time This is the time it takes to electrically switch from one read/write head
to the next before reading can begin for multihead drives. This takes between 1 and 2 ms.
Media Rate The media rate is the speed at which the head reads bits from the surface of
the disk.
Choosing a storage device depends on particular use case. For example, consider choosing
an instance of Amazon EC2 where various choices are available. The choice rests on factors
such as data volume, data frequency or velocity, complexity of computation, and availability
of processed data.