aboutsummaryrefslogtreecommitdiffstats
path: root/docs/testing/user/userguide/14-nsb-operation.rst
blob: d157914a93eacc5c82e374e3393fcc88b4e13ba4 (plain)
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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
.. This work is licensed under a Creative Commons Attribution 4.0 International
.. License.
.. http://creativecommons.org/licenses/by/4.0
.. (c) OPNFV, 2016-2017 Intel Corporation.

Yardstick - NSB Testing - Operation
===================================

Abstract
--------

NSB test configuration and OpenStack setup requirements


OpenStack Network Configuration
-------------------------------

NSB requires certain OpenStack deployment configurations.
For optimal VNF characterization using external traffic generators NSB requires
provider/external networks.


Provider networks
^^^^^^^^^^^^^^^^^

The VNFs require a clear L2 connect to the external network in order to
generate realistic traffic from multiple address ranges and ports.

In order to prevent Neutron from filtering traffic we have to disable Neutron
Port Security. We also disable DHCP on the data ports because we are binding
the ports to DPDK and do not need DHCP addresses.  We also disable gateways
because multiple default gateways can prevent SSH access to the VNF from the
floating IP.  We only want a gateway on the mgmt network

.. code-block:: yaml

    uplink_0:
      cidr: '10.1.0.0/24'
      gateway_ip: 'null'
      port_security_enabled: False
      enable_dhcp: 'false'

Heat Topologies
^^^^^^^^^^^^^^^

By default Heat will attach every node to every Neutron network that is
created. For scale-out tests we do not want to attach every node to every
network.

For each node you can specify which ports are on which network using the
network_ports dictionary.

In this example we have ``TRex xe0 <-> xe0 VNF xe1 <-> xe0 UDP_Replay``

.. code-block:: yaml

      vnf_0:
        floating_ip: true
        placement: "pgrp1"
        network_ports:
          mgmt:
            - mgmt
          uplink_0:
            - xe0
          downlink_0:
            - xe1
      tg_0:
        floating_ip: true
        placement: "pgrp1"
        network_ports:
          mgmt:
            - mgmt
          uplink_0:
            - xe0
          # Trex always needs two ports
          uplink_1:
            - xe1
      tg_1:
        floating_ip: true
        placement: "pgrp1"
        network_ports:
          mgmt:
           - mgmt
          downlink_0:
           - xe0

Collectd KPIs
-------------

NSB can collect KPIs from collected.  We have support for various plugins
enabled by the Barometer project.

The default yardstick-samplevnf has collectd installed. This allows for
collecting KPIs from the VNF.

Collecting KPIs from the NFVi is more complicated and requires manual setup.
We assume that collectd is not installed on the compute nodes.

To collectd KPIs from the NFVi compute nodes:


    * install_collectd on the compute nodes
    * create pod.yaml for the compute nodes
    * enable specific plugins depending on the vswitch and DPDK

    example pod.yaml section for Compute node running collectd.

.. code-block:: yaml

    -
      name: "compute-1"
      role: Compute
      ip: "10.1.2.3"
      user: "root"
      ssh_port: "22"
      password: ""
      collectd:
        interval: 5
        plugins:
          # for libvirtd stats
          virt: {}
          intel_pmu: {}
          ovs_stats:
            # path to OVS socket
            ovs_socket_path: /var/run/openvswitch/db.sock
          intel_rdt: {}



Scale-Up
--------

VNFs performance data with scale-up

  * Helps to figure out optimal number of cores specification in the Virtual
    Machine template creation or VNF
  * Helps in comparison between different VNF vendor offerings
  * Better the scale-up index, indicates the performance scalability of a
    particular solution

Heat
^^^^
For VNF scale-up tests we increase the number for VNF worker threads.  In the
case of VNFs we also need to increase the number of VCPUs and memory allocated
to the VNF.

An example scale-up Heat testcase is:

.. literalinclude:: /submodules/yardstick/samples/vnf_samples/nsut/vfw/tc_heat_rfc2544_ipv4_1rule_1flow_64B_trex_scale-up.yaml
   :language: yaml

This testcase template requires specifying the number of VCPUs, Memory and Ports.
We set the VCPUs and memory using the ``--task-args`` options

.. code-block:: console

  yardstick task start --task-args='{"mem": 10480, "vcpus": 4, "ports": 2}' \
  samples/vnf_samples/nsut/vfw/tc_heat_rfc2544_ipv4_1rule_1flow_64B_trex_scale-up.yaml

In order to support ports scale-up, traffic and topology templates need to be used in testcase.

A example topology template is:

.. literalinclude:: /submodules/yardstick/samples/vnf_samples/nsut/vfw/vfw-tg-topology-scale-up.yaml
   :language: yaml

This template has ``vports`` as an argument. To pass this argument it needs to
be configured in ``extra_args`` scenario definition. Please note that more
argument can be defined in that section. All of them will be passed to topology
and traffic profile templates

For example:

.. code-block:: yaml

   schema: yardstick:task:0.1
   scenarios:
   - type: NSPerf
     traffic_profile: ../../traffic_profiles/ipv4_throughput-scale-up.yaml
     extra_args:
       vports: {{ vports }}
     topology: vfw-tg-topology-scale-up.yaml

A example traffic profile template is:

.. literalinclude:: /submodules/yardstick/samples/vnf_samples/traffic_profiles/ipv4_throughput-scale-up.yaml
   :language: yaml

There is an option to provide predefined config for SampleVNFs. Path to config
file may by specified in ``vnf_config`` scenario section.

.. code-block:: yaml

   vnf__0:
      rules: acl_1rule.yaml
      vnf_config: {lb_config: 'SW', file: vfw_vnf_pipeline_cores_4_ports_2_lb_1_sw.conf }


Baremetal
^^^^^^^^^
  1. Follow above traffic generator section to setup.
  2. Edit num of threads in
     ``<repo>/samples/vnf_samples/nsut/vfw/tc_baremetal_rfc2544_ipv4_1rule_1flow_64B_trex_scale_up.yaml``
     e.g, 6 Threads for given VNF

.. code-block:: yaml


     schema: yardstick:task:0.1
     scenarios:
     {% for worker_thread in [1, 2 ,3 , 4, 5, 6] %}
     - type: NSPerf
       traffic_profile: ../../traffic_profiles/ipv4_throughput.yaml
       topology: vfw-tg-topology.yaml
       nodes:
         tg__0: trafficgen_1.yardstick
         vnf__0: vnf.yardstick
       options:
         framesize:
           uplink: {64B: 100}
           downlink: {64B: 100}
         flow:
           src_ip: [{'tg__0': 'xe0'}]
           dst_ip: [{'tg__0': 'xe1'}]
           count: 1
         traffic_type: 4
         rfc2544:
           allowed_drop_rate: 0.0001 - 0.0001
         vnf__0:
           rules: acl_1rule.yaml
           vnf_config: {lb_config: 'HW', lb_count: 1, worker_config: '1C/1T', worker_threads: {{worker_thread}}}
           nfvi_enable: True
       runner:
         type: Iteration
         iterations: 10
         interval: 35
     {% endfor %}
     context:
       type: Node
       name: yardstick
       nfvi_type: baremetal
       file: /etc/yardstick/nodes/pod.yaml

Scale-Out
--------------------

VNFs performance data with scale-out helps

  * in capacity planning to meet the given network node requirements
  * in comparison between different VNF vendor offerings
  * better the scale-out index, provides the flexibility in meeting future
    capacity requirements


Standalone
^^^^^^^^^^

Scale-out not supported on Baremetal.

1. Follow above traffic generator section to setup.
2. Generate testcase for standalone virtualization using ansible scripts

  .. code-block:: console

    cd <repo>/ansible
    trex: standalone_ovs_scale_out_trex_test.yaml or standalone_sriov_scale_out_trex_test.yaml
    ixia: standalone_ovs_scale_out_ixia_test.yaml or standalone_sriov_scale_out_ixia_test.yaml
    ixia_correlated: standalone_ovs_scale_out_ixia_correlated_test.yaml or standalone_sriov_scale_out_ixia_correlated_test.yaml

  update the ovs_dpdk or sriov above Ansible scripts reflect the setup

3. run the test

  .. code-block:: console

    <repo>/samples/vnf_samples/nsut/tc_sriov_vfw_udp_ixia_correlated_scale_out-1.yaml
    <repo>/samples/vnf_samples/nsut/tc_sriov_vfw_udp_ixia_correlated_scale_out-2.yaml

Heat
^^^^

There are sample scale-out all-VM Heat tests. These tests only use VMs and
don't use external traffic.

The tests use UDP_Replay and correlated traffic.

.. code-block:: console

  <repo>/samples/vnf_samples/nsut/cgnapt/tc_heat_rfc2544_ipv4_1flow_64B_trex_correlated_scale_4.yaml

To run the test you need to increase OpenStack CPU, Memory and Port quotas.


Traffic Generator tuning
------------------------

The TRex traffic generator can be setup to use multiple threads per core, this
is for multiqueue testing.

TRex does not automatically enable multiple threads because we currently cannot
detect the number of queues on a device.

To enable multiple queue set the ``queues_per_port`` value in the TG VNF
options section.

.. code-block:: yaml

  scenarios:
    - type: NSPerf
      nodes:
        tg__0: tg_0.yardstick

      options:
        tg_0:
          queues_per_port: 2


Standalone configuration
------------------------

NSB supports certain Standalone deployment configurations.
Standalone supports provisioning a VM in a standalone visualised environment using kvm/qemu.
There two types of Standalone contexts available: OVS-DPDK and SRIOV.
OVS-DPDK uses OVS network with DPDK drivers.
SRIOV enables network traffic to bypass the software switch layer of the Hyper-V stack.

Standalone with OVS-DPDK
^^^^^^^^^^^^^^^^^^^^^^^^

SampleVNF image is spawned in a VM on a baremetal server.
OVS with DPDK is installed on the baremetal server.

.. note:: Ubuntu 17.10 requires DPDK v.17.05 and higher, DPDK v.17.05 requires OVS v.2.8.0.

Default values for OVS-DPDK:

  * queues: 4
  * lcore_mask: ""
  * pmd_cpu_mask: "0x6"

Sample test case file
^^^^^^^^^^^^^^^^^^^^^

  1. Prepare SampleVNF image and copy it to ``flavor/images``.
  2. Prepare context files for TREX and SampleVNF under ``contexts/file``.
  3. Add bridge named ``br-int`` to the baremetal where SampleVNF image is deployed.
  4. Modify ``networks/phy_port`` accordingly to the baremetal setup.
  5. Run test from:

.. literalinclude:: /submodules/yardstick/samples/vnf_samples/nsut/acl/tc_ovs_rfc2544_ipv4_1rule_1flow_64B_trex.yaml
   :language: yaml