From 29c470d38f795ea8673b614597b3be89816ae2ad Mon Sep 17 00:00:00 2001 From: Vratko Polak Date: Tue, 21 Jan 2020 12:02:46 +0100 Subject: Ndrpdr: Add latency trials for smaller loads Change-Id: I46bd4b58bdd02a3deed38c158214e1e9977dde05 Signed-off-by: Vratko Polak (cherry picked from commit 8840972e7bbc45735f6e82e41ad369728c1b370e) --- resources/libraries/python/Constants.py | 4 + .../robot/performance/performance_utils.robot | 89 +++++++++++++++++++--- 2 files changed, 82 insertions(+), 11 deletions(-) (limited to 'resources/libraries') diff --git a/resources/libraries/python/Constants.py b/resources/libraries/python/Constants.py index eded294f60..4e600098d5 100644 --- a/resources/libraries/python/Constants.py +++ b/resources/libraries/python/Constants.py @@ -217,6 +217,10 @@ class Constants: # Duration of one trial in MRR test. PERF_TRIAL_DURATION = get_float_from_env(u"PERF_TRIAL_DURATION", 1.0) + # Duration of one latency-specific trial in NDRPDR test. + PERF_TRIAL_LATENCY_DURATION = get_float_from_env( + u"PERF_TRIAL_LATENCY_DURATION", 5.0) + # UUID string of DUT1 /tmp volume created outside of the # DUT1 docker in case of vpp-device test. ${EMPTY} value means that # /tmp directory is inside the DUT1 docker. diff --git a/resources/libraries/robot/performance/performance_utils.robot b/resources/libraries/robot/performance/performance_utils.robot index 345a31d8e5..e4454739c0 100644 --- a/resources/libraries/robot/performance/performance_utils.robot +++ b/resources/libraries/robot/performance/performance_utils.robot @@ -53,18 +53,20 @@ | | ... | - doublings - How many doublings to do when expanding [1]. Type: int | | ... | - traffic_directions - Bi- (2) or uni- (1) directional traffic. | | ... | Type: int +| | ... | - latency_duration - Duration for latency-specific trials. Type: float | | | | ... | *Example:* | | | | ... | \| Find NDR and PDR intervals using optimized search \| \${0.005} \ | | ... | \| \${0.005} \| \${30.0} \| \${1.0} \| \${2} \| \${600.0} \| \${2} \ -| | ... | \| \${2} \| +| | ... | \| \${2} \| ${5.0} \| | | | | [Arguments] | ${packet_loss_ratio}=${0.005} | | ... | ${final_relative_width}=${0.005} | ${final_trial_duration}=${30.0} | | ... | ${initial_trial_duration}=${1.0} | | ... | ${number_of_intermediate_phases}=${2} | ${timeout}=${720.0} | | ... | ${doublings}=${2} | ${traffic_directions}=${2} +| | ... | ${latency_duration}=${PERF_TRIAL_LATENCY_DURATION} | | | | ${result} = | Perform optimized ndrpdr search | ${frame_size} | | ... | ${traffic_profile} | ${10000} | ${max_rate} @@ -76,15 +78,42 @@ | | Check NDRPDR interval validity | ${result.pdr_interval} | | ... | ${packet_loss_ratio} | | Check NDRPDR interval validity | ${result.ndr_interval} -| | ${rate_sum}= | Set Variable | ${result.ndr_interval.measured_low.target_tr} -| | ${rate_per_stream}= | Evaluate | ${rate_sum} / float(${traffic_directions}) -| | Send traffic at specified rate -| | ... | ${2.0} | ${rate_per_stream}pps | ${framesize} | ${traffic_profile} -| | ... | traffic_directions=${traffic_directions} -| | ${rate_sum}= | Set Variable | ${result.pdr_interval.measured_low.target_tr} -| | ${rate_per_stream}= | Evaluate | ${rate_sum} / float(${traffic_directions}) +| | # Rate needs to be high enough for latency streams. +| | ${rate}= | Set Variable | ${9500} +| | Measure and show latency at specified rate | Latency at 0%: +| | ... | ${latency_duration} | ${rate}pps | ${framesize} +| | ... | ${traffic_profile} | ${traffic_directions} +| | ${ndr_sum}= | Set Variable | ${result.ndr_interval.measured_low.target_tr} +| | ${ndr_per_stream}= | Evaluate | ${ndr_sum} / float(${traffic_directions}) +| | ${rate}= | Evaluate | 0.1 * ${ndr_per_stream} +| | Measure and show latency at specified rate | Latency at 10% NDR: +| | ... | ${latency_duration} | ${rate}pps | ${framesize} +| | ... | ${traffic_profile} | ${traffic_directions} +| | ${rate}= | Evaluate | 0.5 * ${ndr_per_stream} +| | Measure and show latency at specified rate | Latency at 50% NDR: +| | ... | ${latency_duration} | ${rate}pps | ${framesize} +| | ... | ${traffic_profile} | ${traffic_directions} +| | ${rate}= | Evaluate | 0.9 * ${ndr_per_stream} +| | Measure and show latency at specified rate | Latency at 90% NDR: +| | ... | ${latency_duration} | ${rate}pps | ${framesize} +| | ... | ${traffic_profile} | ${traffic_directions} +| | ${pdr_sum}= | Set Variable | ${result.pdr_interval.measured_low.target_tr} +| | ${pdr_per_stream}= | Evaluate | ${pdr_sum} / float(${traffic_directions}) +| | ${rate}= | Evaluate | 0.1 * ${pdr_per_stream} +| | Measure and show latency at specified rate | Latency at 10% PDR: +| | ... | ${latency_duration} | ${rate}pps | ${framesize} +| | ... | ${traffic_profile} | ${traffic_directions} +| | ${rate}= | Evaluate | 0.5 * ${pdr_per_stream} +| | Measure and show latency at specified rate | Latency at 50% PDR: +| | ... | ${latency_duration} | ${rate}pps | ${framesize} +| | ... | ${traffic_profile} | ${traffic_directions} +| | ${rate}= | Evaluate | 0.9 * ${pdr_per_stream} +| | Measure and show latency at specified rate | Latency at 90% PDR: +| | ... | ${latency_duration} | ${rate}pps | ${framesize} +| | ... | ${traffic_profile} | ${traffic_directions} +| | # Finally, a trial with runtime and other stats. | | Send traffic at specified rate -| | ... | ${2.0} | ${rate_per_stream}pps | ${framesize} | ${traffic_profile} +| | ... | ${1.0} | ${ndr_per_stream}pps | ${framesize} | ${traffic_profile} | | ... | traffic_directions=${traffic_directions} | Find Throughput Using MLRsearch @@ -361,7 +390,9 @@ | Send traffic at specified rate | | [Documentation] -| | ... | Send traffic at specified rate. +| | ... | Perform a warmup, show runtime counters during it. +| | ... | Then send traffic at specified rate, possibly multiple trials. +| | ... | Show various DUT stats, optionally also packet trace. | | ... | Return list of measured receive rates. | | ... | The rate argument should be TRex friendly, so it should include "pps". | | @@ -381,7 +412,7 @@ | | ... | *Example:* | | | | ... | \| Send traffic at specified rate \| \${1.0} \| 4.0mpps \| \${64} \ -| | ... | \| 3-node-IPv4 \| \${10} \| \${2} \| \${0} | \${1} \| \${False} +| | ... | \| 3-node-IPv4 \| \${10} \| \${2} \| \${0} \| \${1} \| \${False} \| | | | | [Arguments] | ${trial_duration} | ${rate} | ${frame_size} | | ... | ${traffic_profile} | ${subsamples}=${1} | ${traffic_directions}=${2} @@ -416,6 +447,42 @@ | | ... | Show Packet Trace On All Duts | ${nodes} | maximum=${100} | | Return From Keyword | ${results} +| Measure and show latency at specified rate +| | [Documentation] +| | ... | Send traffic at specified rate, single trial. +| | ... | Extract latency information and append it to text message. +| | ... | The rate argument should be TRex friendly, so it should include "pps". +| | +| | ... | *Arguments:* +| | ... | - message_prefix - Preface to test message addition. Type: string +| | ... | - trial_duration - Duration of single trial [s]. Type: float +| | ... | - rate - Rate for sending packets. Type: string +| | ... | - frame_size - L2 Frame Size [B]. Type: integer/string +| | ... | - traffic_profile - Name of module defining traffic for measurements. +| | ... | Type: string +| | ... | - traffic_directions - Bi- (2) or uni- (1) directional traffic. +| | ... | Type: int +| | ... | - tx_port - TX port of TG, default 0. Type: integer +| | ... | - rx_port - RX port of TG, default 1. Type: integer +| | +| | ... | *Example:* +| | +| | ... | \| Measure and show latency at specified rate \| Latency at 90% NDR \ +| | ... | \| \${1.0} \| 4.0mpps \| \${64} \| 3-node-IPv4 \| \${2} \| +| | +| | [Arguments] | ${message_prefix} | ${trial_duration} | ${rate} +| | ... | ${frame_size} | ${traffic_profile} | ${traffic_directions}=${2} +| | ... | ${tx_port}=${0} | ${rx_port}=${1} +| | +| | # The following line is skipping some default arguments, +| | # that is why subsequent arguments have to be named. +| | Send traffic on tg | ${trial_duration} | ${rate} | ${frame_size} +| | ... | ${traffic_profile} | warmup_time=${0} +| | ... | traffic_directions=${traffic_directions} | tx_port=${tx_port} +| | ... | rx_port=${rx_port} +| | ${latency} = | Get Latency Int +| | Set Test Message | ${\n}${message_prefix} ${latency} | append=${True} + | Clear and show runtime counters with running traffic | | [Documentation] | | ... | Start traffic at specified rate then clear runtime counters on all -- cgit 1.2.3-korg