Contents | Previous | Next

TCP/IP Stack Performance Improvements

VM's TCP/IP stack can act as a router for other VM TCP/IP stacks, or for guests such as Linux. Enhancements were made to TCP/IP 430 to improve performance of the VM TCP/IP stack by optimizing high-use paths, improving algorithms and implementing performance related features. All improvements made were in the device driver layer. The focus of this work was primarily on the performance of the stack when it is acting as a router. However, it was found that the performance was improved not only for the router case but for the stack in general.

Methodology:   This section summarizes the results of a performance evaluation comparing TCP/IP 420 with TCP/IP 430 for HiperSockets, Guest LAN (HiperSockets virtualization), QDIO GbE on OSA Express (QDIO), CLAW and virtual CTC. An LCS device was also measured but the results are not shown here since there was virtually no difference between 420 and 430.

Measurements were done using 420 CP with APAR VM62938 applied and TCP/IP 420 with APAR PQ51738 applied (HiperSockets enablement) or 430 CP and TCP/IP 430.

An internal tool was used to drive request-response (RR), connect-request-response (CRR) and streaming (S) workloads. The RR workload consisted of the client sending 200 bytes to the server and the server responding with 1000 bytes. The CRR workload had the client connecting, sending 64 bytes to the server, the server responding with 8K and the client then disconnecting. The streaming workload consisted of the client sending 20 bytes to the server and the server responding with 20MB.

Each workload was run using HiperSockets, Guest LAN (HiperSockets virtualization), QDIO, CLAW and VCTC connectivity with different MTU sizes. For HiperSockets and Guest LAN, 56K and 1500 were chosen for the streaming workloads, 1500 for CRR workloads and for RR workloads. For QDIO, 8992 and 1500 were chosen for streaming workloads, and 1500 for CRR and RR workloads. For CLAW and VCTC, 2000 and 1500 respectively were used for all workloads. All measurements included 1, 5 and 10 client-server pairs.

The measurements were done on a 2064-109 in an LPAR with 2 dedicated processors. The LPAR had 1GB central storage and 2GB expanded storage. CP monitor data were captured during the measurement run and reduced using VMPRF.

Figure 1. Environment



Figure stkenv not displayed.


Figure 1 shows the measurement environment. All clients communicated with TCPIP1, which used VCTC to communicate with the router stack (TCPIP2). TCPIP2 communicated with the server stack (TCPIP3) using HiperSockets, Guest LAN (HiperSockets), QDIO, CLAW or VCTC. Communication between the client and TCPIP1 and the server and TCPIP3 was done with IUCV.

The following charts show the comparison between results on TCP/IP 420 and the enhancements on TCP/IP 430. The charts show the ratio between the two releases for router stack (TCPIP2) CPU time, server stack (TCPIP3) CPU time, and throughput. For all these charts, a ratio of 1 signifies equivalence between the two releases. The charts show the ratios for the 10 client case, but the ratios for 1 and 5 clients are similar. Specific details are mentioned, as appropriate, after the charts.

Figure 2. Router Stack CPU Time Ratios - 10 clients



Figure tcp2r not displayed.


The shorter the bar, the larger the gain in efficiency for the router. Larger gains were seen for both HiperSockets and Guest LAN when the MTU size was 1500. With the larger MTU size it makes sense that the benefit would be less since larger packets mean we cross the device driver interface fewer times for the same amount of data. All connectivity types gained for all workloads.

Figure 3. Server Stack CPU Time Ratios - 10 clients



Figure tcp3r not displayed.


The server stack shows the same trend as the router stack by gaining the most with HiperSockets and Guest LAN when the MTU size is 1500. The CRR workload showed the smallest improvement for this stack. The gains, in this case, are not as great because the base costs are higher due to the overhead of managing connect-disconnect.

Figure 4. Throughput Ratios - 10 clients



Figure unitr not displayed.


For this chart, the taller the bar the better. Both HiperSockets and Guest LAN with MTU size of 1500 showed the greatest increase while CRR again shows a small improvement. By looking at the tables later in this section, we can see that the server stack (tcpip3) is the bottleneck, using more than 90% of one CPU.

Figure 5. CPU msec/MB Streaming with HiperSockets



Figure msechs not displayed.


This chart, and the next one, show a comparison for CPU usage for the whole system. These are a sample of the data collected. For full data refer to the tables that follow. In these charts the stacks are broken out with the category "other" including the client and server workload tool.

Since one of the drivers targeted for improvement was VCTC, and TCPIP1 communicated with the router stack using VCTC, an improvement is seen for this stack as well.

Figure 6. CPU msec/trans CRR with HiperSockets



Figure msecch not displayed.


This one is interesting as it shows that the Server stack has very high CPU Time as compared to the other stacks and to "other".

Results:   The results are summarized in the following tables. MB/sec (megabytes per second) or trans/sec (transactions per second) was supplied by the workload driver and shows the throughput rate. All other values are from CP monitor data or derived from CP monitor data.

Total_cpu_util

This field was obtained from the SYSTEM_SUMMARY_BY_TIME VMPRF report that shows the average of both processors out of 100%.

tcpip_tot_cpu_util

This field was calculated from the USER_RESOURCE_UTILIZATION VMPRF report (CPU seconds, total). 100% is the equivalent of one fully utilized processor.

cpu_msec/MB

This field was calculated from the previous tot_cpu_util divided by the number of megabytes per second to show the number of milliseconds of time per megabyte.

cpu_msec/trans

This field was calculated from the previous tot_cpu_util divided by the number of transactions per second to show the number of milliseconds of time per transaction.


Table 1. HiperSockets - Streaming 1500


Number of clients
runid
TCPIP level


01
420hss12
420


01
311hss13
430


01


ratio


05
420hss51
420


05
311hss52
430


05


ratio


10
420hssa3
420


10
311hssa2
430


10


ratio


MB/sec
total_cpu_util


25.58
82.10


40.48
72.10


1.58
0.88


26.54
88.80


40.41
77.80


1.52
0.88


24.80
84.70


38.47
81.50


1.55
0.96


tcpip1_tot_cpu_util
tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


40.61
48.18
56.97


52.12
22.12
48.18


1.28
0.46
0.85


45.15
47.27
59.70


54.55
21.21
50.61


1.21
0.45
0.85


44.55
41.21
57.27


55.76
21.82
50.61


1.25
0.53
0.88


cpu_msec/MB
emul_msec/MB
cp_msec/MB


64.19
49.65
14.54


35.62
23.91
11.71


0.55
0.48
0.81


66.92
51.85
15.07


38.51
26.38
12.13


0.58
0.51
0.80


68.31
52.90
15.40


42.37
29.06
13.31


0.62
0.55
0.86


tcpip1_cpu_msec/MB
tcpip1_vcpu_msec/MB
tcpip1_ccpu_msec/MB


15.87
9.60
6.28


12.88
7.04
5.84


0.81
0.73
0.93


17.01
9.93
7.08


13.50
7.35
6.15


0.79
0.74
0.87


17.96
9.90
8.06


14.49
7.72
6.77


0.81
0.78
0.84


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


18.84
15.16
3.67


5.46
3.67
1.80


0.29
0.24
0.49


17.81
14.73
3.08


5.25
3.67
1.57


0.29
0.25
0.51


16.62
14.54
2.08


5.67
3.78
1.89


0.34
0.26
0.91


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


22.27
19.19
3.08


11.90
9.06
2.84


0.53
0.47
0.92


22.49
19.41
3.08


12.52
9.60
2.92


0.56
0.49
0.95


23.09
19.79
3.30


13.15
10.08
3.07


0.57
0.51
0.93

Note: 2064-109; LPAR with 2 dedicated processors

Table 2. HiperSockets - Streaming 56K


Number of clients
runid
TCPIP level


01
420hsb11
420


01
311hsb13
430


01


ratio


05
420hsb52
420


05
311hsb52
430


05


ratio


10
420hsba3
420


10
311hsba3
430


10


ratio


MB/sec
total_cpu_util


58.51
84.00


63.14
73.60


1.08
0.88


62.51
88.10


72.35
84.30


1.16
0.96


60.34
87.10


71.76
86.00


1.19
0.99


tcpip1_tot_cpu_util
tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


51.21
36.67
51.52


49.39
24.55
41.82


0.96
0.67
0.81


56.36
32.42
57.88


57.27
26.36
49.39


1.02
0.81
0.85


55.45
30.61
57.58


58.18
25.76
51.52


1.05
0.84
0.89


cpu_msec/MB
emul_msec/MB
cp_msec/MB


28.71
16.95
11.76


23.31
12.23
11.09


0.81
0.72
0.94


28.19
16.61
11.58


23.30
12.30
11.00


0.83
0.74
0.95


28.87
17.14
11.73


23.97
12.85
11.12


0.83
0.75
0.95


tcpip1_cpu_msec/MB
tcpip1_vcpu_msec/MB
tcpip1_ccpu_msec/MB


8.75
2.90
5.85


7.82
2.54
5.28


0.89
0.88
0.90


9.02
2.76
6.25


7.92
2.51
5.40


0.88
0.91
0.86


9.19
2.81
6.38


8.11
2.62
5.49


0.88
0.93
0.86


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


6.27
4.40
1.86


3.89
2.06
1.82


0.62
0.47
0.98


5.19
3.98
1.21


3.64
2.01
1.63


0.70
0.51
1.35


5.07
3.97
1.10


3.59
2.07
1.52


0.71
0.52
1.38


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


8.80
5.90
2.90


6.62
3.79
2.83


0.75
0.64
0.98


9.26
6.16
3.10


6.83
3.94
2.89


0.74
0.64
0.93


9.54
6.38
3.16


7.18
4.22
2.96


0.75
0.66
0.94

Note: 2064-109; LPAR with 2 dedicated processors

Table 3. HiperSockets - CRR 1500


Number of clients
runid
TCPIP level


01
420hcs12
420


01
311hcs13
430


01


ratio


05
420hcs53
420


05
311hcs52
430


05


ratio


10
420hcsa1
420


10
311hcsa2
430


10


ratio


trans/sec
total_cpu_util


60.84
57.60


63.61
55.90


1.05
0.97


70.28
59.40


75.31
57.30


1.07
0.96


69.87
59.20


73.15
57.30


1.05
0.97


tcpip1_tot_cpu_util
tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


6.67
7.58
89.39


5.45
3.33
91.21


0.82
0.44
1.02


6.36
4.85
94.55


4.85
1.82
95.76


0.76
0.38
1.01


6.06
3.64
94.24


4.85
1.52
95.15


0.80
0.42
1.01


cpu_msec/trans
emul_msec/trans
cp_msec/trans


18.93
17.42
1.51


17.58
16.13
1.45


0.93
0.93
0.96


16.90
15.74
1.17


15.22
14.23
0.98


0.90
0.90
0.84


16.95
15.83
1.12


15.67
14.71
0.96


0.92
0.93
0.86


tcpip1_cpu_msec/trans
tcpip1_vcpu_msec/trans
tcpip1_ccpu_msec/trans


1.10
0.65
0.45


0.86
0.43
0.43


0.78
0.66
0.96


0.91
0.52
0.39


0.64
0.32
0.32


0.70
0.62
0.82


0.87
0.52
0.35


0.66
0.37
0.29


0.76
0.71
0.83


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


1.25
0.95
0.30


0.52
0.24
0.29


0.42
0.25
0.97


0.69
0.52
0.17


0.24
0.12
0.12


0.35
0.23
0.71


0.52
0.43
0.09


0.21
0.12
0.08


0.40
0.28
0.89


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


14.69
14.49
0.20


14.34
14.15
0.19


0.98
0.98
0.95


13.45
13.28
0.17


12.72
12.55
0.16


0.95
0.95
0.94


13.49
13.31
0.17


13.01
12.84
0.17


0.96
0.96
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 4. HiperSockets - RR 1500


Number of clients
runid
TCPIP level


01
420hrs13
420


01
311hrs11
430


01


ratio


05
420hrs51
420


05
311hrs52
430


05


ratio


10
420hrsa2
420


10
311hrsa2
430


10


ratio


trans/sec
total_cpu_util


912.16
80.50


1234.67
72.40


1.35
0.90


1426.26
98.10


1917.20
99.10


1.34
1.01


1577.28
96.80


1993.94
98.40


1.26
1.02


tcpip1_tot_cpu_util
tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


29.39
39.09
41.21


33.33
21.21
25.45


1.13
0.54
0.62


36.97
38.18
47.58


42.73
23.64
35.15


1.16
0.62
0.74


35.76
30.61
45.15


40.61
19.09
34.85


1.14
0.62
0.77


cpu_msec/trans
emul_msec/trans
cp_msec/trans


1.77
1.29
0.47


1.17
0.73
0.44


0.66
0.57
0.94


1.38
1.01
0.36


1.03
0.67
0.36


0.75
0.66
1.00


1.23
0.91
0.32


0.99
0.66
0.33


0.80
0.73
1.03


tcpip1_cpu_msec/trans
tcpip1_vcpu_msec/trans
tcpip1_ccpu_msec/trans


0.32
0.16
0.16


0.27
0.11
0.16


0.84
0.69
1.00


0.26
0.13
0.13


0.22
0.09
0.13


0.85
0.69
1.00


0.23
0.11
0.11


0.20
0.09
0.12


0.87
0.82
1.09


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


0.43
0.33
0.10


0.17
0.08
0.09


0.40
0.24
0.90


0.27
0.20
0.06


0.12
0.06
0.07


0.44
0.30
1.17


0.19
0.15
0.04


0.10
0.05
0.05


0.53
0.33
1.25


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


0.45
0.36
0.10


0.21
0.12
0.09


0.47
0.33
0.90


0.33
0.25
0.08


0.18
0.10
0.08


0.55
0.40
1.00


0.29
0.20
0.08


0.17
0.10
0.08


0.59
0.50
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 5. Guest LAN HIPER - Streaming 1500


Number of clients
runid
TCPIP level


01
420gss13
420


01
311gss12
430


01


ratio


05
420gss52
420


05
311gss53
430


05


ratio


10
420gssa2
420


10
311gssa3
430


10


ratio


MB/sec
total_cpu_util


23.66
83.90


39.31
80.30


1.66
0.96


24.82
93.80


44.11
94.60


1.78
1.01


23.77
93.10


40.44
93.70


1.70
1.01


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


47.88
65.45


26.97
62.42


0.56
0.95


48.48
71.52


28.48
72.73


0.59
1.02


45.15
69.09


26.36
70.00


0.58
1.01


cpu_msec/MB
emul_msec/MB
cp_msec/MB


70.92
49.62
21.30


40.85
23.00
17.86


0.58
0.46
0.84


75.58
53.02
22.56


42.89
24.80
18.09


0.57
0.47
0.80


78.33
55.11
23.22


46.34
27.25
19.09


0.59
0.49
0.82


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


20.24
15.63
4.61


6.86
4.09
2.78


0.34
0.26
0.60


19.53
15.26
4.27


6.46
3.98
2.47


0.33
0.26
0.58


19.00
15.30
3.70


6.52
4.05
2.47


0.34
0.26
0.67


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


27.66
18.70
8.97


15.88
7.86
8.02


0.57
0.42
0.89


28.81
19.29
9.52


16.49
8.38
8.11


0.57
0.43
0.85


29.07
19.38
9.69


17.31
8.84
8.47


0.60
0.46
0.87

Note: 2064-109; LPAR with 2 dedicated processors

Table 6. Guest LAN HIPER - Streaming 56K


Number of clients
runid
TCPIP level


01
420gsb11
420


01
311gsb13
430


01


ratio


05
420gsb52
420


05
311gsb51
430


05


ratio


10
420gsba3
420


10
311gsba3
430


10


ratio


MB/sec
total_cpu_util


53.34
84.40


54.81
64.70


1.03
0.77


59.93
88.00


70.06
85.00


1.17
0.97


57.25
87.20


71.06
88.10


1.24
1.01


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


41.52
52.73


24.24
40.00


0.58
0.76


35.45
58.48


29.39
53.33


0.83
0.91


33.33
58.79


29.39
56.06


0.88
0.95


cpu_msec/MB
emul_msec/MB
cp_msec/MB


31.65
17.44
14.21


23.61
11.17
12.44


0.75
0.64
0.88


29.37
15.85
13.52


24.26
11.56
12.70


0.83
0.73
0.94


30.46
16.63
13.83


24.80
11.96
12.83


0.81
0.72
0.93


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


7.78
5.17
2.61


4.42
2.54
1.88


0.57
0.49
0.72


5.92
4.60
1.31


4.20
2.55
1.64


0.71
0.55
1.25


5.82
4.61
1.22


4.14
2.60
1.54


0.71
0.56
1.26


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


9.89
5.06
4.83


7.30
2.71
4.59


0.74
0.54
0.95


9.76
4.85
4.90


7.61
2.90
4.71


0.78
0.60
0.96


10.27
5.19
5.08


7.89
3.07
4.82


0.77
0.59
0.95

Note: 2064-109; LPAR with 2 dedicated processors

Table 7. Guest LAN HIPER - CRR 1500


Number of clients
runid
TCPIP level


01
420gcs12
420


01
311gcs12
430


01


ratio


05
420gcs51
420


05
311gcs52
430


05


ratio


10
420gcsa2
420


10
311gcsa2
430


10


ratio


trans/sec
total_cpu_util


63.05
57.20


65.99
55.30


1.05
0.97


69.56
59.20


72.12
57.50


1.04
0.97


70.54
58.20


70.99
57.40


1.01
0.99


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


9.09
83.03


4.24
84.24


0.47
1.01


5.15
93.94


2.42
94.55


0.47
1.01


3.64
94.55


1.82
94.55


0.50
1.00


cpu_msec/trans
emul_msec/trans
cp_msec/trans


18.14
16.37
1.78


16.76
15.09
1.67


0.92
0.92
0.94


17.02
15.73
1.29


15.95
14.75
1.19


0.94
0.94
0.92


16.50
15.37
1.13


16.17
15.04
1.13


0.98
0.98
1.00


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


1.44
1.01
0.43


0.64
0.28
0.37


0.44
0.28
0.86


0.74
0.52
0.22


0.34
0.13
0.21


0.46
0.25
0.95


0.52
0.39
0.13


0.26
0.13
0.13


0.50
0.33
1.00


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


13.17
12.83
0.34


12.77
12.40
0.37


0.97
0.97
1.09


13.50
13.20
0.30


13.11
12.86
0.25


0.97
0.97
0.83


13.40
13.15
0.26


13.32
13.06
0.26


0.99
0.99
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 8. Guest LAN HIPER - RR 1500


Number of clients
runid
TCPIP level


01
420grs11
420


01
311grs11
430


01


ratio


05
420grs52
420


05
311grs52
430


05


ratio


10
420grsa2
420


10
311grsa2
430


10


ratio


trans/sec
total_cpu_util


677.26
57.50


958.57
52.30


1.42
0.91


1414.92
97.50


1866.59
99.00


1.32
1.02


1547.37
97.00


1922.05
98.60


1.24
1.02


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


30.61
31.21


18.18
21.82


0.59
0.70


39.39
48.48


26.06
38.18


0.66
0.79


32.73
46.67


22.12
37.58


0.68
0.81


cpu_msec/trans
emul_msec/trans
cp_msec/trans


1.70
1.21
0.48


1.09
0.65
0.44


0.64
0.54
0.92


1.38
0.98
0.40


1.06
0.66
0.40


0.77
0.67
1.00


1.25
0.90
0.35


1.03
0.66
0.37


0.82
0.73
1.06


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


0.45
0.33
0.13


0.19
0.08
0.11


0.42
0.24
0.85


0.28
0.19
0.08


0.14
0.05
0.09


0.50
0.26
1.13


0.21
0.15
0.06


0.12
0.04
0.07


0.57
0.27
1.17


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


0.46
0.34
0.12


0.23
0.12
0.11


0.50
0.35
0.92


0.34
0.23
0.11


0.20
0.10
0.11


0.59
0.43
1.00


0.30
0.20
0.10


0.20
0.09
0.10


0.67
0.45
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 9. QDIO - Streaming 1500


Number of clients
runid
TCPIP level


01
420qns12
420


01
311qns13
430


01


ratio


05
420qns52
420


05
311qns52
430


05


ratio


10
420qnsa2
420


10
311qnsa3
430


10


ratio


MB/sec
total_cpu_util


26.38
70.60


34.10
69.50


1.29
0.98


30.64
85.70


40.55
84.90


1.32
0.99


29.48
87.40


39.05
87.90


1.32
1.01


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


35.15
43.94


26.97
40.61


0.77
0.92


38.48
52.12


28.79
46.97


0.75
0.90


37.67
52.33


27.88
48.48


0.74
0.93


cpu_msec/MB
emul_msec/MB
cp_msec/MB


53.53
38.74
14.78


40.76
26.28
14.49


0.76
0.68
0.98


55.94
40.60
15.34


41.87
27.47
14.40


0.75
0.68
0.94


59.29
43.08
16.21


45.02
29.65
15.36


0.76
0.69
0.95


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


13.33
10.11
3.22


7.91
4.27
3.64


0.59
0.42
1.13


12.56
9.89
2.67


7.10
3.81
3.29


0.57
0.39
1.23


12.78
10.06
2.71


7.14
3.96
3.18


0.56
0.39
1.17


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


16.66
13.55
3.10


11.91
8.71
3.20


0.71
0.64
1.03


17.01
13.85
3.16


11.58
8.59
2.99


0.68
0.62
0.95


17.75
14.36
3.39


12.42
9.16
3.26


0.70
0.64
0.96

Note: 2064-109; LPAR with 2 dedicated processors

Table 10. QDIO - Streaming 8992


Number of clients
runid
TCPIP level


01
420qjs12
420


01
311qjs12
430


01


ratio


05
420qjs53
420


05
311qjs51
430


05


ratio


10
420qjsa2
420


10
311qjsa1
430


10


ratio


MB/sec
total_cpu_util


46.53
80.20


60.16
81.50


1.29
1.02


53.15
89.20


66.67
88.70


1.25
0.99


49.99
87.20


61.89
87.60


1.24
1.00


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


38.18
43.94


31.21
40.00


0.82
0.91


38.18
51.21


29.39
47.88


0.77
0.93


35.15
51.82


26.97
50.91


0.77
0.98


cpu_msec/MB
emul_msec/MB
cp_msec/MB


34.47
20.89
13.58


27.09
14.30
12.80


0.79
0.68
0.94


33.57
20.13
13.43


26.61
14.25
12.36


0.79
0.71
0.92


34.89
21.08
13.80


28.31
15.54
12.76


0.81
0.74
0.92


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


8.21
5.54
2.67


5.19
2.47
2.72


0.63
0.45
1.02


7.18
5.25
1.94


4.41
2.45
1.95


0.61
0.47
1.01


7.03
5.33
1.70


4.36
2.69
1.66


0.62
0.50
0.98


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


9.44
6.45
3.00


6.65
3.73
2.92


0.70
0.58
0.97


9.64
6.33
3.31


7.18
4.05
3.14


0.74
0.64
0.95


10.37
6.61
3.76


8.23
4.60
3.62


0.79
0.70
0.96

Note: 2064-109; LPAR with 2 dedicated processors

Table 11. QDIO CRR 1500


Number of clients
runid
TCPIP level


01
420qnc13
420


01
311qnc11
430


01


ratio


05
420qnc52
420


05
311qnc52
430


05


ratio


10
420qnca2
420


10
311qnca3
430


10


ratio


trans/sec
total_cpu_util


60.72
53.10


61.27
52.20


1.01
0.98


71.26
57.60


70.91
57.20


1.00
0.99


70.69
57.00


70.99
56.80


1.00
1.00


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


7.27
80.91


3.33
84.24


0.46
1.04


3.94
93.33


2.12
94.24


0.54
1.01


2.42
93.64


1.52
93.94


0.63
1.00


cpu_msec/trans
emul_msec/trans
cp_msec/trans


17.49
15.94
1.55


17.04
15.54
1.50


0.97
0.97
0.97


16.17
15.10
1.07


16.13
15.03
1.10


1.00
1.00
1.03


16.13
15.14
0.99


16.00
14.99
1.01


0.99
0.99
1.02


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


1.20
0.90
0.30


0.54
0.25
0.30


0.45
0.28
1.00


0.55
0.43
0.13


0.30
0.13
0.17


0.55
0.30
1.31


0.34
0.26
0.09


0.21
0.13
0.09


0.62
0.50
1.00


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


13.32
13.08
0.25


13.75
13.55
0.20


1.03
1.04
0.80


13.10
12.88
0.21


13.29
13.08
0.21


1.01
1.02
1.00


13.25
13.03
0.21


13.23
13.02
0.21


1.00
1.00
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 12. QDIO - RR 1500


Number of clients
runid
TCPIP level


01
420qnr12
420


01
311qnr13
430


01


ratio


05
420qnr51
420


05
311qnr51
430


05


ratio


10
420qnra2
420


10
311qnra2
430


10


ratio


trans/sec
total_cpu_util


661.69
54.50


817.55
48.40


1.24
0.89


1423.79
95.30


1820.44
95.90


1.28
1.01


1591.39
92.20


1944.78
94.90


1.22
1.03


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


25.76
27.88


14.55
16.97


0.56
0.61


34.85
45.45


20.91
34.55


0.60
0.76


24.55
40.91


16.06
33.64


0.65
0.82


cpu_msec/trans
emul_msec/trans
cp_msec/trans


1.65
1.19
0.46


1.18
0.72
0.46


0.72
0.61
1.00


1.34
0.98
0.36


1.05
0.69
0.37


0.78
0.70
1.03


1.16
0.85
0.31


0.98
0.65
0.32


0.84
0.76
1.03


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


0.39
0.29
0.10


0.18
0.08
0.10


0.46
0.28
1.00


0.24
0.18
0.06


0.11
0.05
0.06


0.46
0.28
1.00


0.15
0.12
0.04


0.08
0.04
0.04


0.53
0.33
1.00


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


0.42
0.33
0.10


0.21
0.12
0.09


0.50
0.36
0.90


0.32
0.23
0.09


0.19
0.10
0.09


0.59
0.43
1.00


0.26
0.17
0.08


0.17
0.09
0.08


0.65
0.53
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 13. CLAW - Streaming 2000


Number of clients
runid
TCPIP level


01
420css13
420


01
208css13
430


01


ratio


05
420css52
420


05
208css53
430


05


ratio


10
420cssa2
420


10
208cssa3
430


10


ratio


MB/sec
total_cpu_util


2.13
8.90


2.17
7.00


1.02
0.79


2.28
9.90


2.34
8.00


1.03
0.81


2.21
10.80


2.26
9.10


1.02
0.84


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


4.85
3.33


3.03
2.73


0.62
0.82


4.85
3.64


3.03
3.03


0.62
0.83


4.74
3.86


2.98
3.33


0.63
0.86


cpu_msec/MB
emul_msec/MB
cp_msec/MB


83.57
53.52
30.05


64.52
38.71
25.81


0.77
0.72
0.86


86.84
57.02
29.82


68.38
41.88
26.50


0.79
0.73
0.89


97.74
64.25
33.48


80.53
50.44
30.09


0.82
0.79
0.90


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


22.76
12.80
9.96


13.96
5.59
8.38


0.61
0.44
0.84


21.27
11.96
9.30


12.95
5.18
7.77


0.61
0.43
0.84


21.43
11.91
9.53


13.20
5.43
7.76


0.62
0.46
0.81


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


15.65
11.38
4.27


12.57
8.38
4.19


0.80
0.74
0.98


15.95
11.96
3.99


12.95
9.07
3.89


0.81
0.76
0.97


17.46
12.70
4.76


14.75
10.09
4.66


0.84
0.79
0.98

Note: 2064-109; LPAR with 2 dedicated processors

Table 14. CLAW - CRR 2000


Number of clients
runid
TCPIP level


01
420ccs12
420


01
208ccs12
430


01


ratio


05
420ccs52
420


05
208ccs52
430


05


ratio


10
420ccsa3
420


10
208ccsa3
430


10


ratio


trans/sec
total_cpu_util


43.01
34.00


44.09
32.30


1.03
0.95


62.09
51.60


62.75
51.00


1.01
0.99


65.46
54.40


66.27
53.60


1.01
0.99


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


4.55
47.27


3.33
46.36


0.73
0.98


4.55
78.48


3.03
79.39


0.67
1.01


3.33
85.76


2.42
85.76


0.73
1.00


cpu_msec/trans
emul_msec/trans
cp_msec/trans


15.81
13.67
2.14


14.65
12.57
2.09


0.93
0.92
0.98


16.62
15.01
1.61


16.25
14.66
1.59


0.98
0.98
0.99


16.62
15.28
1.34


16.18
14.85
1.33


0.97
0.97
0.99


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


1.06
0.56
0.49


0.76
0.27
0.48


0.72
0.48
0.98


0.73
0.39
0.34


0.48
0.19
0.29


0.66
0.49
0.85


0.51
0.32
0.19


0.37
0.14
0.23


0.73
0.44
1.21


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


10.99
10.71
0.28


10.52
10.24
0.27


0.96
0.96
0.96


12.64
12.40
0.24


12.65
12.41
0.24


1.00
1.00
1.00


13.10
12.87
0.23


12.94
12.71
0.23


0.99
0.99
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 15. CLAW - RR 2000


Number of clients
runid
TCPIP level


01
420crs11
420


01
208crs11
430


01


ratio


05
420crs52
420


05
208crs51
430


05


ratio


10
420crsa2
420


10
208crsa3
430


10


ratio


trans/sec
total_cpu_util


258.85
17.40


264.54
15.50


1.02
0.89


461.62
25.80


468.94
22.80


1.02
0.88


494.71
24.80


504.52
23.10


1.02
0.93


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


6.67
6.33


5.15
5.45


0.77
0.86


5.76
9.09


3.94
7.88


0.68
0.87


3.33
8.79


2.12
7.88


0.64
0.90


cpu_msec/trans
emul_msec/trans
cp_msec/trans


1.34
0.83
0.51


1.17
0.67
0.50


0.87
0.81
0.98


1.12
0.76
0.36


0.97
0.63
0.34


0.87
0.83
0.94


1.00
0.72
0.29


0.92
0.63
0.28


0.92
0.88
0.97


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


0.26
0.14
0.12


0.19
0.07
0.13


0.73
0.50
1.08


0.12
0.07
0.05


0.08
0.03
0.05


0.67
0.43
1.00


0.07
0.04
0.02


0.04
0.02
0.02


0.57
0.50
1.00


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


0.24
0.14
0.10


0.21
0.10
0.10


0.88
0.71
1.00


0.20
0.11
0.09


0.17
0.08
0.08


0.85
0.73
0.89


0.18
0.10
0.08


0.16
0.07
0.08


0.89
0.70
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Table 16. VCTC - Streaming 1500


Number of clients
runid
TCPIP level


01
420vss13
420


01
208vss12
430


01


ratio


05
420vss51
420


05
208vss52
430


05


ratio


10
420vssa2
420


10
208vssa1
430


10


ratio


MB/sec
total_cpu_util


31.00
80.80


35.43
79.50


1.14
0.98


33.85
95.30


40.33
95.40


1.19
1.00


32.20
96.70


39.70
96.40


1.23
1.00


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


37.88
52.73


32.12
53.64


0.85
1.02


38.79
59.70


34.85
60.91


0.90
1.02


37.33
59.67


33.94
59.70


0.91
1.00


cpu_msec/MB
emul_msec/MB
cp_msec/MB


52.13
31.61
20.52


44.88
25.35
19.53


0.86
0.80
0.95


56.31
34.92
21.39


47.31
27.47
19.84


0.84
0.79
0.93


60.06
37.52
22.55


48.56
28.56
20.00


0.81
0.76
0.89


tcpip2_cpu_msec/MB
tcpip2_vcpu_msec/MB
tcpip2_ccpu_msec/MB


12.22
6.65
5.57


9.07
3.42
5.64


0.74
0.51
1.01


11.46
6.62
4.83


8.64
3.38
5.26


0.75
0.51
1.09


11.59
6.73
4.87


8.55
3.36
5.19


0.74
0.50
1.07


tcpip3_cpu_msec/MB
tcpip3_vcpu_msec/MB
tcpip3_ccpu_msec/MB


17.01
9.87
7.14


15.14
8.13
7.01


0.89
0.82
0.98


17.64
10.47
7.16


15.10
8.34
6.76


0.86
0.80
0.94


18.53
10.97
7.56


15.04
8.40
6.64


0.81
0.77
0.88

Note: 2064-109; LPAR with 2 dedicated processors

Table 17. VCTC - CRR 1500


Number of clients
runid
TCPIP level


01
420vcs12
420


01
208vcs12
430


01


ratio


05
420vcs52
420


05
208vcs52
430


05


ratio


10
420vcsa1
420


10
208vcsa2
430


10


ratio


trans/sec
total_cpu_util


61.07
56.60


66.42
56.10


1.09
0.99


70.91
59.50


72.47
58.90


1.02
0.99


69.33
58.50


70.72
58.60


1.02
1.00


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


6.06
84.24


4.85
78.18


0.80
0.93


3.94
94.24


2.42
94.24


0.61
1.00


3.03
93.33


1.82
94.55


0.60
1.01


cpu_msec/trans
emul_msec/trans
cp_msec/trans


18.54
16.47
2.06


16.89
14.91
1.99


0.91
0.91
0.97


16.78
15.29
1.49


16.26
14.99
1.27


0.97
0.98
0.85


16.88
15.52
1.36


16.57
15.41
1.16


0.98
0.99
0.85


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


0.99
0.45
0.55


0.73
0.23
0.50


0.74
0.51
0.91


0.56
0.30
0.26


0.33
0.13
0.21


0.59
0.43
0.81


0.44
0.22
0.22


0.26
0.09
0.17


0.59
0.41
0.77


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


13.79
13.25
0.55


11.77
11.27
0.50


0.85
0.85
0.91


13.29
12.91
0.38


13.00
12.67
0.33


0.98
0.98
0.87


13.46
13.11
0.35


13.37
13.07
0.30


0.99
1.00
0.86

Note: 2064-109; LPAR with 2 dedicated processors

Table 18. VCTC - RR 1500


Number of clients
runid
TCPIP level


01
420vrs12
420


01
208vrs13
430


01


ratio


05
420vrs51
420


05
208vrs53
430


05


ratio


10
420vrsa2
420


10
208vrsa1
430


10


ratio


trans/sec
total_cpu_util


811.40
53.30


908.05
51.00


1.12
0.96


1604.08
98.90


1792.13
99.00


1.12
1.00


1680.30
98.80


1866.83
99.00


1.11
1.00


tcpip2_tot_cpu_util
tcpip3_tot_cpu_util


22.42
24.85


18.79
23.03


0.84
0.93


32.73
42.73


27.88
40.61


0.85
0.95


29.39
41.82


24.85
39.70


0.85
0.95


cpu_msec/trans
emul_msec/trans
cp_msec/trans


1.31
0.79
0.52


1.12
0.62
0.50


0.85
0.78
0.96


1.23
0.77
0.47


1.10
0.65
0.46


0.89
0.84
0.98


1.18
0.75
0.43


1.06
0.64
0.42


0.90
0.85
0.98


tcpip2_cpu_msec/trans
tcpip2_vcpu_msec/trans
tcpip2_ccpu_msec/trans


0.28
0.13
0.15


0.21
0.06
0.14


0.75
0.46
0.93


0.20
0.09
0.11


0.16
0.04
0.11


0.80
0.44
1.00


0.17
0.08
0.10


0.13
0.04
0.09


0.76
0.50
0.90


tcpip3_cpu_msec/trans
tcpip3_vcpu_msec/trans
tcpip3_ccpu_msec/trans


0.31
0.16
0.15


0.25
0.11
0.14


0.81
0.69
0.93


0.27
0.13
0.14


0.23
0.09
0.14


0.85
0.69
1.00


0.25
0.12
0.13


0.21
0.09
0.13


0.84
0.75
1.00

Note: 2064-109; LPAR with 2 dedicated processors

Contents | Previous | Next