Chapter 2. Hardware configuration planning 33
.
Figure 2-11 Read intensive - 15K rpm versus 10K rpm drives
Figure 2-11 illustrates the example when a read intensive workload was run on both the 10K
rpm and the 15K rpm configurations. The 15K rpm configuration performs better, always
delivering lower response times for all throughputs as compared to the 10K rpm configuration.
You can appreciate comparing the results in Figure 2-10 on page 32 and Figure 2-11, that
when running a higher cache hit ratio workload, the performance gain is less significant, than
when running the lower cache hit ratio workload.
For more detailed estimation, see 4.1, “Disk Magic” on page 86 and Chapter 12,
“Understanding your workload” on page 407.
2.8 RAID implementation
In this section we describe the characteristics of the RAID implementation in the DS6000.
The performance considerations regarding the possible RAID configurations are discussed in
detail later in Chapter 3, “Logical configuration planning” on page 53.
2.8.1 RAID Arrays
The basic unit where data is stored in the DS6000 is the disk drive module (also referred to as
a DDM). Disk drives for the DS6000 are available in capacities of 73 GB, 146 GB, 300 GB
(capacity and speed), and are grouped together in an
four-pack, and these four-packs can be
installed in each on the FC-AL loops. One FC-AL loop can hold up to 16 four-packs, which
means that the maximum number of 64 DDMs can be found in a loop. The
raw or physical
capacity
installed on a loop will depend on the four-packs (their capacity) that are installed on
that loop. Physical capacity is discussed in 2.6.3, “Disk four-pack capacity” on page 27.
Logically, four DDMs are grouped in an
Array Site automatically. As illustrated in Figure 2-12,
initially four DDMs of four 4-packs (16 DDMs) are randomly selected and make up the
Array
Site
. After initial setup, the next Array Site is configured by adding more four-pack DDMs.
Note: These results are just from the simulation of Disk Magic, your performance may
change according to your environment.
Read intensive workload
0
2
4
6
8
10
12
14
16
18
20
0 200040006000800010000
Throughput(IOPS)
response time(ms)
146GB10K Resp time
146GB15K Resp time
read:write=7:3
cache hit ratio=28%
I/O block size=4kB
write efficiency=33%