docs/testing/user/userguide/opnfv_yardstick_tc011.rst

   1 .. This work is licensed under a Creative Commons Attribution 4.0 International
   2 .. License.
   3 .. http://creativecommons.org/licenses/by/4.0
   4 .. (c) OPNFV, Huawei Technologies Co.,Ltd and others.
   5
   6 *************************************
   7 Yardstick Test Case Description TC011
   8 *************************************
   9
  10 .. _iperf3: https://iperf.fr/
  11
  12 +-----------------------------------------------------------------------------+
  13 |Packet delay variation between VMs                                           |
  14 |                                                                             |
  15 +--------------+--------------------------------------------------------------+
  16 |test case id  | OPNFV_YARDSTICK_TC011_PACKET DELAY VARIATION BETWEEN VMs     |
  17 |              |                                                              |
  18 +--------------+--------------------------------------------------------------+
  19 |metric        | jitter: packet delay variation (ms)                          |
  20 |              |                                                              |
  21 +--------------+--------------------------------------------------------------+
  22 |test purpose  | The purpose of TC011 is to evaluate the IaaS network         |
  23 |              | performance with regards to network jitter (packet delay     |
  24 |              | variation).                                                  |
  25 |              | It measures the packet delay variation sending the packets   |
  26 |              | from one VM to the other.                                    |
  27 |              |                                                              |
  28 |              | The purpose is also to be able to spot the trends.           |
  29 |              | Test results, graphs and similar shall be stored for         |
  30 |              | comparison reasons and product evolution understanding       |
  31 |              | between different OPNFV versions and/or configurations.      |
  32 |              |                                                              |
  33 +--------------+--------------------------------------------------------------+
  34 |test tool     | iperf3                                                       |
  35 |              |                                                              |
  36 |              | iPerf3 is a tool for active measurements of the maximum      |
  37 |              | achievable bandwidth on IP networks. It supports tuning of   |
  38 |              | various parameters related to timing, buffers and protocols. |
  39 |              | The UDP protocols can be used to measure jitter delay.       |
  40 |              |                                                              |
  41 |              | (iperf3 is not always part of a Linux distribution, hence it |
  42 |              | needs to be installed. It is part of the Yardstick Docker    |
  43 |              | image. As an example see the /yardstick/tools/ directory for |
  44 |              | how to generate a Linux image with pktgen included.)         |
  45 |              |                                                              |
  46 +--------------+--------------------------------------------------------------+
  47 |test          | iperf3 test is invoked between a host VM and a target VM.    |
  48 |description   |                                                              |
  49 |              | Jitter calculations are continuously computed by the server, |
  50 |              | as specified by RTP in RFC 1889. The client records a 64 bit |
  51 |              | second/microsecond timestamp in the packet. The server       |
  52 |              | computes the relative transit time as (server's receive time |
  53 |              | - client's send time). The client's and server's clocks do   |
  54 |              | not need to be synchronized; any difference is subtracted    |
  55 |              | outin the jitter calculation. Jitter is the smoothed mean of |
  56 |              | differences between consecutive transit times.               |
  57 |              |                                                              |
  58 +--------------+--------------------------------------------------------------+
  59 |configuration | File: opnfv_yardstick_tc011.yaml                             |
  60 |              |                                                              |
  61 |              | * options:                                                   |
  62 |              |   protocol: udp # The protocol used by iperf3 tools          |
  63 |              |   # Send the given number of packets without pausing:        |
  64 |              |   bandwidth: 20m                                             |
  65 |              | * runner:                                                    |
  66 |              |   duration: 30 # Total test duration 30 seconds.             |
  67 |              |                                                              |
  68 |              | * SLA (optional):                                            |
  69 |              |   jitter: 10 (ms) # The maximum amount of jitter that is     |
  70 |              |   accepted.                                                  |
  71 |              |                                                              |
  72 +--------------+--------------------------------------------------------------+
  73 |applicability | Test can be configured with different:                       |
  74 |              |                                                              |
  75 |              | * bandwidth: Test case can be configured with different      |
  76 |              |              bandwidth.                                      |
  77 |              |                                                              |
  78 |              | * duration: The test duration can be configured.             |
  79 |              |                                                              |
  80 |              | * jitter: SLA is optional. The SLA in this test case         |
  81 |              |           serves as an example.                              |
  82 |              |                                                              |
  83 +--------------+--------------------------------------------------------------+
  84 |usability     | This test case is one of Yardstick's generic test. Thus it   |
  85 |              | is runnable on most of the scenarios.                        |
  86 |              |                                                              |
  87 +--------------+--------------------------------------------------------------+
  88 |references    | iperf3_                                                      |
  89 |              |                                                              |
  90 |              | ETSI-NFV-TST001                                              |
  91 |              |                                                              |
  92 +--------------+--------------------------------------------------------------+
  93 |pre-test      | The test case image needs to be installed into Glance        |
  94 |conditions    | with iperf3 included in the image.                           |
  95 |              |                                                              |
  96 |              | No POD specific requirements have been identified.           |
  97 |              |                                                              |
  98 +--------------+--------------------------------------------------------------+
  99 |test sequence | description and expected result                              |
 100 |              |                                                              |
 101 +--------------+--------------------------------------------------------------+
 102 |step 1        | Two host VMs with iperf3 installed are booted, as server and |
 103 |              | client.                                                      |
 104 |              |                                                              |
 105 +--------------+--------------------------------------------------------------+
 106 |step 2        | Yardstick is connected with the host VM by using ssh.        |
 107 |              | A iperf3 server is started on the server VM via the ssh      |
 108 |              | tunnel.                                                      |
 109 |              |                                                              |
 110 +--------------+--------------------------------------------------------------+
 111 |step 3        | iperf3 benchmark is invoked. Jitter is calculated and check  |
 112 |              | against the SLA. Logs are produced and stored.               |
 113 |              |                                                              |
 114 |              | Result: Logs are stored.                                     |
 115 |              |                                                              |
 116 +--------------+--------------------------------------------------------------+
 117 |step 4        | The host VMs are deleted.                                    |
 118 |              |                                                              |
 119 +--------------+--------------------------------------------------------------+
 120 |test verdict  | Test should not PASS if any jitter is above the optional SLA |
 121 |              | value, or if there is a test case execution problem.         |
 122 |              |                                                              |
 123 +--------------+--------------------------------------------------------------+