Specifications

to the desired cylinder of the disk. The latency is the amount of time it takes for
the disk to rotate to the proper sector on that cylinder.
It should be noted that two disks of the same physical size, for example 3.5-inch
disks, will differ in their access times with the larger capacity disk having a
better access time. This is due to the fact that the distance between cylinders is
shorter on the larger disk and, therefore, seek time is reduced. This is the
primary reason that disk access times have been reduced as capacities have
been increased.
Maximum transfer rate is the rate at which the device can deliver data back to
the SCSI adapter. It mainly depends on the processor/DMA controller integrated
on the device but can be no more than the SCSI maximum data transfer rate, for
example 20 MBps for a SCSI-II Fast/Wide interface.
Caching is important for the same reason it is important on other subsystems;
namely, it speeds up the time it takes to perform routine operations. For high
performance, the drive should be able to provide write caching. With write
caching, the drive signals the completion of the write immediately after it
receives the data but before the data is actually written to disk. The system then
continues to do other work while the hard disk is actually writing the data.
Performance is significantly better because subsequent disk operations can be
overlapped with this cached write operation.
The following table summarizes the specifications on current IBM PC Server
hard disks:
Table 5. Summary of Disks Performance Characteristics
Disk Average
Seek
Time
Burst
Transfer
Rate
Sustained
Transfer
Rate
Average
Latency
MTBF
(K hours)
1GB Fast 8.6 ms 10 MBps 3.7-4.5 MBps 5.56 ms 500
2GB Fast 9.5 ms 10 MBps 3.7-4.5 MBps 5.56 ms 500
1.08GB Fast 10.5 ms 10 MBps 3.2-4.0 MBps 5.56 ms 500
1.12GB F/W 6.9 ms 20 MBps 5.5-7.4 MBps 4.17 ms 1000
2.25GB F/W 7.5 ms 20 MBps 5.5-7.4 MBps 4.17 ms 1000
4.51GB F/W 8.0 ms 20 MBps 5.5-7.4 MBps 4.17 ms 1000
1.6.5 RAID Technology
Several factors have contributed to the growing popularity of disk arrays:
•
Performance
The capacity of single large disks has grown rapidly, but the performance
improvements have been modest, when compared to the advances made in
the other subsystems that make up a computer system. The reason for this
is that disks are mechanical devices, affected by delays in positioning
read/write heads (seeks) and the rotation time of the media (Latency).
•
Reliability
22 NetWare Integration Guide