Performance Disclaimer: The measurements presented here were taken using a configuration that was also being used for other projects. Although we strove to have repeatable and comparable results, the measurements were not taken in a controlled environment. Actual results might vary. Performance information is provided “AS IS” and no warranties or guarantees are expressed or implied by IBM. For official IBM performance information, refer to the Large Systems Performance Reference documentation.
|
Note: Because the link type pays such a large role in sysplex performance, in this chapter we use the full name of each link type (for example, HCA3-O 12X IFB3) in every case to ensure that there is no confusion or misunderstanding.
|
CF/Host
|
z10 BC
|
z10 EC
|
z114
|
z196
|
zBC12
|
zEC12
|
z10 EC ISC3
|
16%
|
17%
|
17%
|
21%
|
19%
|
24%
|
z10 EC ICB4
|
9%
|
10%
|
NA
|
NA
|
NA
|
NA
|
z10 EC 12X IFB
|
11%
|
12%
|
12%
|
14%
|
14%
|
16%
|
z196 ISC3
|
16%
|
17%
|
17%
|
21%
|
19%
|
24%
|
z196 12X IFB
|
11%
|
12%
|
11%
|
14%
|
14%
|
15%
|
z196 12X IFB3
|
NA
|
NA
|
9%
|
11%
|
10%
|
12%
|
The XES Synch/Async heuristic algorithm effectively caps overhead at about 18%.
These values are based on nine CF requests per second per MIPS.
|
Important: The results are reported here cannot be used to determine service times to be seen in your configuration. There are many variables that can affect service times. It is possible that your service times will be better than those that we observed, or they might be longer. What is more valuable is to compare the service time we got from one configuration with the service times we observed in a different configuration.
|
Structure type and attributes
|
Target request rate
|
IMS Shared Message Queue
|
1000
|
DB2 GBP 4 KB heavy write bias
|
2500
|
DB2 GBP 32 KB heavy read bias
|
2500
|
DB2 GBP 4 KB heavy read bias
|
2500
|
DB2 GBP 4 KB read bias
|
2500
|
DB2 GBP 4 KB read bias
|
2500
|
DB2 Lock
|
10000
|
GRS Lock
|
5000
|
IMS Lock
|
5000
|
IMS Cache 4 KB read bias
|
3000
|
IMS Cache 4 KB read bias
|
3000
|
System Logger 4 K only writes
|
1000
|
System Logger 1KB mainly writes
|
1000
|
WebSphere MQ 1 KB read and write
|
500
|
WebSphere MQ 1 KB read and write
|
500
|
WebSphere MQ 1 KB read and write
|
500
|
WebSphere MQ 1 KB read and write
|
500
|
WebSphere MQ 1 KB read and write
|
500
|
WebSphere MQ 1 KB read and write
|
500
|
Total (per z/OS system)
|
44,500
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
ICB4
|
2
|
2
|
z10
|
24.3%
|
94.6%
|
14.1
|
5.4%
|
84
|
86959
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
|
ICB4
|
2
|
2
|
z10
|
24.3%
|
94.6
|
14.1
|
5.4
|
84
|
86959
|
|
HCA2-O
|
2
|
8
|
z10
|
36.7%
|
90.3
|
18.7
|
9.7
|
92.4
|
85435
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
ICB4
|
2
|
2
|
z10
|
24.3%
|
94.6
|
14.1
|
5.4
|
84
|
86959
|
HCA2-O
|
2
|
8
|
z10
|
36.7%
|
90.3
|
18.7
|
9.7
|
92.4
|
85435
|
HCA2-O to z196
|
2
|
8
|
z196
|
34.9%1
|
95.4
|
14
|
4.6
|
105.5
|
86204
|
Note: All the measurements in this section were taken using a 1-way z196 CF. The reported CPU utilizations were not normalized to the z10 measurements, however the text takes note of this where appropriate.
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
ISC3
|
8
|
8
|
z196
|
52.1%
|
41.2
|
26.5
|
58.8
|
105.5
|
76528
|
Note: The design intent of Parallel Sysplex is that if a CF is unavailable, its workload can be moved to another CF in the sysplex. Therefore, CFs should be sized so that each CF has sufficient capacity to be able to run the entire CF workload and still deliver acceptable service times. The general IBM guideline is that the combined CF CPU utilization is to not exceed 70% to 80%, meaning that if you have two CFs and a balanced configuration, each CF should not exceed about 35% to 40% busy.
In the example here, if you have two CFs and both of them are over 50% utilized, the utilization of the surviving CF in case of an outage would be over 100%, which is well above the IBM guideline.
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
|
ISC3
|
8
|
8
|
z196
|
52.1%
|
41.2
|
26.5
|
58.8
|
105.5
|
76528
|
|
HCA2-O
|
2
|
8
|
z196
|
34.5%
|
94.7
|
14
|
5.3
|
108.3
|
87175
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
|
ISC3
|
8
|
8
|
z196
|
52.1%
|
41.2
|
26.5
|
58.8
|
105.5
|
76528
|
|
HCA2-O
|
2
|
8
|
z196
|
34.5%
|
94.7
|
14
|
5.3
|
108.3
|
87175
|
|
HCA3-O
|
2
|
8
|
z196
|
25.8%
|
99.7%
|
7
|
0.3
|
148.5
|
89476
|
Note: This set of measurements was for HCA3-O 12X links, with HCA3-O adapters at both ends of the link, and with only four CHPIDs assigned to those ports, meaning that the ports were running in IFB3 mode.
If the HCA3-O 12X ports had been connected to HCA2-O 12X ports, or had more than four CHPIDs assigned to them, they would have run in IFB mode and the performance would have been equivalent to the performance of HCA2-O 12X links.
|
Note: All the measurements in this section have been normalized to the capacity of our z196 and the request rate that was achieved during the HCA3-O 12X IFB3 measurement.
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
ISC3
|
8
|
8
|
z196
|
52.1%
|
41.2
|
26.5
|
58.8
|
105.5
|
76528
|
Link
type
|
Number
links
|
Number
CHPIDs
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
|
ISC3
|
8
|
8
|
z196
|
52.1%
|
41.2
|
26.5
|
58.8
|
105.5
|
76528
|
|
HCA2-O LR
|
2
|
8
|
z196
|
34.7%
|
83.7
|
19.5
|
16.3
|
99.7
|
84099
|
Link
type
|
#
links
|
#
CHPIDs
|
Total # subch
|
CF type
|
CF %
util
|
Sync %
|
Sync serv time
|
Async %
|
Async serv time
|
Total req rate
|
|
ISC3
|
8
|
8
|
56
|
z196
|
52.1%
|
41.2
|
26.5
|
58.8
|
105.5
|
76528
|
|
HCA2-O LR
|
2
|
8
|
56
|
z196
|
34.7%
|
83.7
|
19.5
|
16.3
|
99.7
|
84099
|
|
HCA3-O LR
|
2
|
8
|
256
|
z196
|
34.9%
|
82.01
|
20
|
17.99
|
105.5
|
81388
|
Link
type
|
Number
links to each CF
|
Number
CHPIDs to each CF
|
CF types
|
Total req rate
|
ICB4
|
2
|
2
|
z10
|
75607
|
Link
type/
CF CPU type
|
CF1 %
Util
|
CF1 Sync %
|
CF1 Sync serv time
|
CF1 Async %
|
CF1 Async serv time
|
CF2 % Util
|
CF2 Sync %
|
CF2 Sync serv time
|
CF2 Async %
|
CF2 Async serv time
|
ICB4
z10
|
33.2%
|
71.24
|
17
|
28.76
|
73.5
|
16.8%
|
0.44
|
42
|
99.56
|
72
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
|
ICB4 z10
|
46.4
|
42
|
10022
|
76
|
46
|
44
|
10022
|
74
|
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
|
ICB4 z10
|
29.7
|
36
|
7199
|
71
|
29.7
|
38
|
7199
|
68
|
|
Link
type
|
Number
links to each CF
|
Number
CHPIDs to each CF
|
CF types
|
Total req rate
|
ICB4
|
2
|
2
|
z10
|
75607
|
HCA2-O
|
2
|
8
|
z10
|
71810
|
Link
type/
CF CPU type
|
CF1 %
Util
|
CF1 Sync %
|
CF1 Sync serv time
|
CF1 Async %
|
CF1 Async serv time
|
CF2 % Util
|
CF2 Sync %
|
CF2 Sync serv time
|
CF2 Async %
|
CF2 Async serv time
|
ICB4
z10
|
33.2%
|
71.24
|
17
|
28.76
|
73.5
|
16.8%
|
0.44
|
42
|
99.56
|
72
|
HCA2-O
z10
|
44.2%
|
67.3
|
21.5
|
32.7
|
104.2
|
16.4%
|
0.40
|
78.25
|
99.6
|
114
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
ICB4 z10
|
46.4
|
42
|
10022
|
76
|
46
|
44
|
10022
|
74
|
HCA2 z10
|
33.5
|
81.3
|
8073
|
116.5
|
33.8
|
82.1
|
8073
|
117.8
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
ICB4 z10
|
29.7
|
36
|
7199
|
71
|
29.7
|
38
|
7199
|
68
|
HCA2 z10
|
25.2
|
71.4
|
6195
|
109.6
|
25.2
|
73.1
|
6195
|
109.0
|
Link
type
|
Number
links to each CF
|
Number
CHPIDs to each CF
|
CF types
|
Total req rate
|
ICB4
|
2
|
2
|
z10
|
75607
|
HCA2-O
|
2
|
8
|
z10
|
71810
|
HCA2-O on z196
|
2
|
8
|
z196
|
74514
|
Link
type/
CF CPU type
|
CF1 %
Util
|
CF1 Sync %
|
CF1 Sync serv time
|
CF1 Async %
|
CF1 Async serv time
|
CF2 % Util
|
CF2 Sync %
|
CF2 Sync serv time
|
CF2 Async %
|
CF2 Async serv time
|
|
ICB4
z10
|
33.2%
|
71.24
|
17
|
28.76
|
73.5
|
16.8%
|
0.44
|
42
|
99.56
|
72
|
|
HCA2-O z10
|
44.2%
|
67.3
|
21.5
|
32.7
|
104.2
|
16.4%
|
0.40
|
78.25
|
99.6
|
114
|
|
HCA2-O z196
|
43.1%
|
72.4
|
18
|
27.6
|
97
|
17.1%
|
0.41
|
65
|
99.59
|
102
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
|
ICB4 z10
|
46.4
|
42
|
10022
|
76
|
46
|
44
|
10022
|
74
|
|
HCA2 z10
|
33.5
|
81.3
|
8073
|
116.5
|
33.8
|
82.1
|
8073
|
117.8
|
|
HCA2 z196
|
37.9
|
69
|
9085
|
101
|
37.6
|
69
|
9085
|
105
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
|
ICB4 z10
|
29.7
|
36
|
7199
|
71
|
29.7
|
38
|
7199
|
68
|
|
HCA2 z10
|
25.2
|
71.4
|
6195
|
109.6
|
25.2
|
73.1
|
6195
|
109.0
|
|
HCA2 z196
|
27.3
|
60
|
6758
|
92
|
27.3
|
59
|
6758
|
97
|
Link
type
|
Number
links to each CF
|
Number
CHPIDs to each CF
|
CF types
|
Total req rate
|
ICB4
|
2
|
2
|
z10
|
75607
|
HCA2-O
|
2
|
8
|
z10
|
71810
|
HCA2-O on z196
|
2
|
8
|
z196
|
74514
|
HCA3-O IFB3 on z196
|
2
|
8
|
z196
|
77467
|
Link
type/
CF CPU type
|
CF1 %
Util
|
CF1 Sync %
|
CF1 Sync serv time
|
CF1 Async %
|
CF1 Async serv time
|
CF2 % Util
|
CF2 Sync %
|
CF2 Sync serv time
|
CF2 Async %
|
CF2 Async serv time
|
|
ICB4
z10
|
33.2%
|
71.24
|
17
|
28.76
|
73.5
|
16.8%
|
0.44
|
42
|
99.56
|
72
|
|
HCA2-O
z10
|
44.2%
|
67.3
|
21.5
|
32.7
|
104.2
|
16.4%
|
0.40
|
78.25
|
99.6
|
114
|
|
HCA2-O z196
|
43.1%
|
72.4
|
18
|
27.6
|
97
|
17.1%
|
0.41
|
65
|
99.59
|
102
|
|
HCA3-O IFB3 z196
|
34.7%
|
76.25
|
9
|
23.75
|
70
|
15.7%
|
0.42
|
37
|
99.58
|
74
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
|
ICB4 z10
|
46.4
|
42
|
10022
|
76
|
46
|
44
|
10022
|
74
|
|
HCA2 z10
|
33.5
|
81.3
|
8073
|
116.5
|
33.8
|
82.1
|
8073
|
117.8
|
|
HCA2 z196
|
37.9
|
69
|
9085
|
101
|
37.6
|
69
|
9085
|
105
|
|
HCA3 IFB3 z196
|
45.3
|
37
|
10283
|
71
|
44.9
|
39
|
10283
|
76
|
Config
|
Prim Sync rate
|
Prim Sync resp time
|
Prim Async rate
|
Prim Async resp time
|
Secondry Sync rate
|
Secondry Sync resp time
|
Secondry Async rate
|
Secondry Async resp time
|
|
ICB4 z10
|
29.7
|
36
|
7199
|
71
|
29.7
|
38
|
7199
|
68
|
|
HCA2 z10
|
25.2
|
71.4
|
6195
|
109.6
|
25.2
|
73.1
|
6195
|
109.0
|
|
HCA2 z196
|
27.3
|
60
|
6758
|
92
|
27.3
|
59
|
6758
|
97
|
|
HCA3 IFB3 z196
|
29.3
|
33
|
7126
|
66
|
29.2
|
34
|
7126
|
71
|