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
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
|
Release Notes
=============
Changes in CSIT |release|
-------------------------
#. **VPP performance tests**
- *MRR tests* - New Maximum Receive Rate tests measure the packet
forwarding rate under the maximum load offered by traffic
generator over a set trial duration, regardless of packet loss.
MRR tests are used for continuous performance trending and for
comparison between releases.
- *Service Chaining with SRv6* - New SRv6 (Segment Routing IPv6) proxy
tests measure performance of SRv6 Endpoint fronting SR-unaware
appliance via masquerading (End.AM), dynamic proxy (End.AD) or
static proxy (End.AS) SR functions.
#. **Presentation and Analytics Layer**
- *Performance trending* - Added continuous performance trending and
analysis. New Performance Trending and Performance Analysis jobs
executed regular throughput tests, with results being subsequently
analysed and trend and anomalies summarized and presented in VPP
Performance Dashboard and trendline graphs.
#. **Test Framework Optimizations**
- *Performance tests efficiency* - Qemu build/install optimizations,
warmup phase handling, vpp restart handling. Resulted in improved
stability and reduced total execution time by 30% for single pkt
size e.g. 64B/78B.
- *General code housekeeping* - ongoing RF keywords optimizations,
removal of redundant RF keywords.
Performance Changes
-------------------
Relative performance changes in measured packet throughput in CSIT
|release| are calculated against the results from CSIT |release-1|
report. Listed mean and standard deviation values are computed based on
a series of the same tests executed against respective VPP releases to
verify test results repeatibility, with percentage change calculated for
mean values. Note that the standard deviation is quite high for a small
number of packet throughput tests, what indicates poor test results
repeatability and makes the relative change of mean throughput value not
fully representative for these tests. The root causes behind poor
results repeatibility vary between the test cases.
NDR Changes
~~~~~~~~~~~
NDR small packet throughput changes between releases are available in a
CSV and pretty ASCII formats:
- `csv format for 1t1c <../_static/vpp/performance-changes-ndr-1t1c-full.csv>`_,
- `csv format for 2t2c <../_static/vpp/performance-changes-ndr-2t2c-full.csv>`_,
- `pretty ASCII format for 1t1c <../_static/vpp/performance-changes-ndr-1t1c-full.txt>`_,
- `pretty ASCII format for 2t2c <../_static/vpp/performance-changes-ndr-2t2c-full.txt>`_.
.. note::
Test results have been generated by
`FD.io test executor vpp performance jobs`_ with Robot Framework result
files csit-vpp-perf-|srelease|-\*.zip `archived here <../_static/archive/>`_.
PDR Changes
~~~~~~~~~~~
NDR small packet throughput changes between releases are available in a
CSV and pretty ASCII formats:
- `csv format for 1t1c <../_static/vpp/performance-changes-pdr-1t1c-full.csv>`_,
- `csv format for 2t2c <../_static/vpp/performance-changes-pdr-2t2c-full.csv>`_,
- `pretty ASCII format for 1t1c <../_static/vpp/performance-changes-pdr-1t1c-full.txt>`_,
- `pretty ASCII format for 2t2c <../_static/vpp/performance-changes-pdr-2t2c-full.txt>`_.
.. note::
Test results have been generated by
`FD.io test executor vpp performance jobs`_ with Robot Framework result
files csit-vpp-perf-|srelease|-\*.zip `archived here <../_static/archive/>`_.
MRR Changes
~~~~~~~~~~~
MRR small packet throughput changes between releases are available in a
CSV and pretty ASCII formats:
- `csv format for 1t1c <../_static/vpp/performance-changes-mrr-1t1c-full.csv>`_,
- `csv format for 2t2c <../_static/vpp/performance-changes-mrr-2t2c-full.csv>`_,
- `csv format for 4t4c <../_static/vpp/performance-changes-mrr-4t4c-full.csv>`_,
- `pretty ASCII format for 1t1c <../_static/vpp/performance-changes-mrr-1t1c-full.txt>`_,
- `pretty ASCII format for 2t2c <../_static/vpp/performance-changes-mrr-2t2c-full.txt>`_,
- `pretty ASCII format for 4t4c <../_static/vpp/performance-changes-mrr-4t4c-full.txt>`_.
.. note::
Test results have been generated by
`FD.io test executor vpp mrr jobs <https://jenkins.fd.io/view/csit/job/csit-vpp-perf-mrr-daily-master/>`_
with Robot Framework result files csit-vpp-perf-mrr-daily-master__*__output.xml.gz
`archived here <https://docs.fd.io/csit/master/trending/_static/archive/>`_.
Comparison Across Testbeds
--------------------------
.. warning::
TODO: Add:
Table 1.
Test Case 3-Node Hsw 3-Node Skx Skx vs. Hsw Delta [%]
Table 2.
Test Case 3-Node Skx 2-Node Skx 2-Node vs. 3-Node Delta [%]
Throughput Trending
-------------------
In addition to reporting throughput changes between VPP releases, CSIT
provides continuous performance trending for VPP master branch:
#. `VPP Performance Dashboard <https://docs.fd.io/csit/master/trending/introduction/index.html>`_
- per VPP test case throughput trend, trend compliance and summary of
detected anomalies.
#. `Trending Methodology <https://docs.fd.io/csit/master/trending/methodology/index.html>`_
- throughput test metrics, trend calculations and anomaly
classification (progression, regression, outlier).
#. `Trendline Graphs <https://docs.fd.io/csit/master/trending/trending/index.html>`_
- per VPP build MRR throughput measurements against the trendline
with anomaly highlights, with associated CSIT test jobs.
Known Issues
------------
List of known issues in CSIT |release| for VPP performance tests:
+---+-------------------------------------------------+------------+-----------------------------------------------------------------+
| # | Issue | Jira ID | Description |
+===+=================================================+============+=================================================================+
| 1 | Sporadic (1 in 200) NDR discovery test failures | CSIT-570 | DPDK reporting rx-errors, indicating L1 issue. Suspected issue |
| | on x520. | | with HW combination of X710-X520 in LF testbeds. Not observed |
| | | | outside of LF testbeds. |
+---+-------------------------------------------------+------------+-----------------------------------------------------------------+
| 2 | Lower than expected NDR throughput of DPDK | CSIT-571 | Suspected NIC firmware or DPDK driver issue affecting NDR and |
| | testpmd and VPP L2 path NDR throughput with | | PDR throughput on XL710 and X710 NICs. |
| | xl710 and x710 NICs, compared to x520 NICs. | | |
+---+-------------------------------------------------+------------+-----------------------------------------------------------------+
| 3 | Tagged Ethernet dot1q and dot1ad L2 path | CSIT-1066 | Tagged Ethernet dot1q and dot1ad L2 path throughput regression: |
| | throughput regression. | | NDR -2%..-5%, PDR -2%..-6%, MRR. Affects l2xc and l2bd |
| | | | performance tests. |
+---+-------------------------------------------------+------------+-----------------------------------------------------------------+
| 4 | IPSec (software, no QAT HW) throughput | CSIT-1064 | IPSec throughput regression: NDR -3%..-8%, PDR -2%..-8%, MRR |
| | regression. | | -3%..-7%. Affects IPSec SW tests, QAT HW tests not affected. |
+---+-------------------------------------------------+------------+-----------------------------------------------------------------+
| 5 | High failure rate of creating working container | CSIT-1065 | About 20% of orchestrated container topology tests failing data |
| | topologies with K8s/Ligato orchestration. | | plane verification indicating configuration issue. Suspected |
| | | | issue with Ligato vpp-agent. |
+---+-------------------------------------------------+------------+-----------------------------------------------------------------+
|
