HP ProLiant 4500 Disk Subsystem Performance and Scalability - Page 20

Like Capacity Scaling, Like Capacity Test Results

Page 20 highlights

ECG025.0997 WHITE PAPER (cont.) ... Like Capacity Scaling Since like capacity scaling can affect your system, it is important to understand the impact it might have on system performance. To be able to determine this information, we tested like capacity scalability by maintaining the same total disk capacity for each test (8GB or 24GB) and added different quantities of drives to a single disk controller. The results of these tests determine if system performance improves when using multiple lower capacity drives or fewer larger capacity drives. In the next few sections of this white paper we analyze these different configurations. Like Capacity Test Results Earlier we discussed concurrency and how the more spindles (disks) you have in your system the better the performance would be because more I/O requests are being concurrently processed. Overall our single disk controller like capacity tests provide evidence that support our theory on concurrency. For example, if you need 8 Gigabytes of storage capacity, our test show the benefits of using four 2GB disks instead of two 4GB disks. The storage capacity is the same; however, the performance increase is 68% as shown in Figure 9. We needed only one 9GB drive in our test to reach the eight Gigabyte storage capacity requirement, so consequently concurrency could not take place and is therefore not beneficial in this configuration. Like Capacity Scaling (RAID 0 - 1 Controller) 16,000,000 Server Throughput (Bytes/sec) 14,000,000 12,000,000 10,000,000 4 x 2GB 2 x 4GB 1 x 9GB 8,000,000 6,000,000 68% 4,000,000 2,000,000 0 4 8 16 20 24 28 32 36 40 44 48 52 56 57 Number of Clients Figure 9: Like Capacity Scaling in a RAID 0 Environment. 20

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
  • 28
  • 29
  • 30

W
HITE
P
APER
(cont.)
20
ECG025.0997
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Like Capacity Scaling
Since like capacity scaling can affect your system, it is important to understand the impact
it might have on system performance.
To be able to determine this information, we tested
like capacity scalability by maintaining the same total disk capacity for each test (8GB or
24GB) and added different quantities of drives to a single disk controller.
The results of
these tests determine if system performance improves when using multiple lower capacity
drives or fewer larger capacity drives.
In the next few sections of this white paper we
analyze these different configurations.
Like Capacity Test Results
Earlier we discussed concurrency and how the more spindles (disks) you have in your
system the better the performance would be because more I/O requests are being
concurrently processed.
Overall our single disk controller like capacity tests provide
evidence that support our theory on concurrency.
For example, if you need 8 Gigabytes
of storage capacity, our test show the benefits of using four 2GB disks instead of two 4GB
disks.
The storage capacity is the same; however, the performance increase is 68% as
shown in Figure 9.
We needed only one 9GB drive in our test to reach the eight Gigabyte
storage capacity requirement, so consequently concurrency could not take place and is
therefore not beneficial in this configuration.
Like Capacity Scaling (RAID 0 - 1 Controller)
0
2,000,000
4,000,000
6,000,000
8,000,000
10,000,000
12,000,000
14,000,000
16,000,000
4
8
16
20
24
28
32
36
40
44
48
52
56
57
Number of Clients
Server Throughput (Bytes/sec)
4 x 2GB
2 x 4GB
1 x 9GB
68%
Figure 9: Like Capacity Scaling in a RAID 0 Environment.