CSIT-102: Add latency measurement to performance testing

- Add latency stream as a background stream in both directions
- Latency background stream is not using the VM transformation engine.
  Raw stream with single packet is created.
- Latency background stream has 1kpps rate with packet of same size as
  traffic stream.
- Display latency values (min/avg/max) in results of TC and reporting
  remaining data including histogram and jitter inside of the
  search/pass keyword.

Change-Id: I78ce4659b57caab08d5729f51a1e14d518fd3273
Signed-off-by: Peter Mikus <pmikus@cisco.com>
Signed-off-by: pmikus <pmikus@cisco.com>

4 files changed, 177 insertions, 72 deletions

diff --git a/resources/libraries/python/DropRateSearch.py b/resources/libraries/python/DropRateSearch.py
index 1f8e5618fe..b0f15b25db 100644
--- a/resources/libraries/python/DropRateSearch.py
+++ b/resources/libraries/python/DropRateSearch.py
@@ -453,14 +453,14 @@ class DropRateSearch(object):
     def verify_search_result(self):
         """Fail if search was not successful.
 
-        :return: Result rate.
-        :rtype: float
+        :return: Result rate and latency stats.
+        :rtype: tuple
         """
         if self._search_result == SearchResults.FAILURE:
             raise Exception('Search FAILED')
         elif self._search_result in [SearchResults.SUCCESS,
                                      SearchResults.SUSPICIOUS]:
-            return self._search_result_rate
+            return self._search_result_rate, self._latency_stats
 
     def binary_search(self, b_min, b_max, traffic_type, skip_max_rate=False):
         """Binary search of rate with loss below acceptance criteria.
diff --git a/resources/libraries/python/TrafficGenerator.py b/resources/libraries/python/TrafficGenerator.py
index 91a43fb0ca..33ff597c8e 100644
--- a/resources/libraries/python/TrafficGenerator.py
+++ b/resources/libraries/python/TrafficGenerator.py
@@ -46,7 +46,10 @@ class TGDropRateSearchImpl(DropRateSearch):
             unit_rate = str(rate) + self.get_rate_type_str()
             tg_instance.trex_stl_start_remote_exec(self.get_duration(),
                                                    unit_rate, frame_size,
-                                                   traffic_type, False)
+                                                   traffic_type)
+            # Get latency stats from stream
+            self._latency_stats = tg_instance.get_latency()
+
             loss = tg_instance.get_loss()
             sent = tg_instance.get_sent()
             if self.loss_acceptance_type_is_percentage():
@@ -74,6 +77,7 @@ class TrafficGenerator(object):
         self._result = None
         self._loss = None
         self._sent = None
+        self._latency = None
         self._received = None
         self._node = None
         # T-REX interface order mapping
@@ -103,6 +107,14 @@ class TrafficGenerator(object):
         """
         return self._received
 
+    def get_latency(self):
+        """Return min/avg/max latency.
+
+        :return: Latency stats.
+        :rtype: list
+        """
+        return self._latency
+
     #pylint: disable=too-many-arguments, too-many-locals
     def initialize_traffic_generator(self, tg_node, tg_if1, tg_if2,
                                      tg_if1_adj_node, tg_if1_adj_if,
@@ -273,7 +285,8 @@ class TrafficGenerator(object):
             raise RuntimeError('T-rex stateless runtime error')
 
     def trex_stl_start_remote_exec(self, duration, rate, framesize,
-                                   traffic_type, async_call, warmup_time=5):
+                                   traffic_type, async_call=False,
+                                   latency=True, warmup_time=5):
         """Execute script on remote node over ssh to start traffic.
 
         :param duration: Time expresed in seconds for how long to send traffic.
@@ -281,12 +294,14 @@ class TrafficGenerator(object):
         :param framesize: L2 frame size to send (without padding and IPG).
         :param traffic_type: Traffic profile.
         :param async_call: If enabled then don't wait for all incomming trafic.
+        :param latency: With latency measurement.
         :param warmup_time: Warmup time period.
         :type duration: int
         :type rate: str
         :type framesize: int
         :type traffic_type: str
         :type async_call: bool
+        :type latency: bool
         :type warmup_time: int
         :return: Nothing
         """
@@ -295,10 +310,9 @@ class TrafficGenerator(object):
 
         _p0 = 1
         _p1 = 2
-        _async = ""
+        _async = "--async" if async_call else ""
+        _latency = "--latency" if latency else ""
 
-        if async_call:
-            _async = "--async"
         if self._ifaces_reordered != 0:
             _p0, _p1 = _p1, _p0
 
@@ -312,9 +326,10 @@ class TrafficGenerator(object):
                 "--p{5}_src_start_ip 20.20.20.1 "
                 "--p{5}_src_end_ip 20.20.20.254 "
                 "--p{5}_dst_start_ip 10.10.10.1 "
-                "{6} --warmup_time={7}'".format(Constants.REMOTE_FW_DIR,
-                                                duration, rate, framesize, _p0,
-                                                _p1, _async, warmup_time),
+                "{6} {7} --warmup_time={8}'".format(Constants.REMOTE_FW_DIR,
+                                                    duration, rate, framesize,
+                                                    _p0, _p1, _async, _latency,
+                                                    warmup_time),
                 timeout=int(duration)+60)
         elif traffic_type in ["3-node-IPv4"]:
             (ret, stdout, stderr) = ssh.exec_command(
@@ -326,9 +341,10 @@ class TrafficGenerator(object):
                 "--p{5}_src_start_ip 20.20.20.2 "
                 "--p{5}_src_end_ip 20.20.20.254 "
                 "--p{5}_dst_start_ip 10.10.10.2 "
-                "{6} --warmup_time={7}'".format(Constants.REMOTE_FW_DIR,
-                                                duration, rate, framesize, _p0,
-                                                _p1, _async, warmup_time),
+                "{6} {7} --warmup_time={8}'".format(Constants.REMOTE_FW_DIR,
+                                                    duration, rate, framesize,
+                                                    _p0, _p1, _async, _latency,
+                                                    warmup_time),
                 timeout=int(duration)+60)
         elif traffic_type in ["3-node-IPv6"]:
             (ret, stdout, stderr) = ssh.exec_command(
@@ -340,9 +356,10 @@ class TrafficGenerator(object):
                 "--p{5}_src_start_ip 2001:2::2 "
                 "--p{5}_src_end_ip 2001:2::FE "
                 "--p{5}_dst_start_ip 2001:1::2 "
-                "{6} --warmup_time={7}'".format(Constants.REMOTE_FW_DIR,
-                                                duration, rate, framesize, _p0,
-                                                _p1, _async, warmup_time),
+                "{6} {7} --warmup_time={8}'".format(Constants.REMOTE_FW_DIR,
+                                                    duration, rate, framesize,
+                                                    _p0, _p1, _async, _latency,
+                                                    warmup_time),
                 timeout=int(duration)+60)
         else:
             raise NotImplementedError('Unsupported traffic type')
@@ -358,6 +375,7 @@ class TrafficGenerator(object):
             self._received = None
             self._sent = None
             self._loss = None
+            self._latency = None
         else:
             # last line from console output
             line = stdout.splitlines()[-1]
@@ -368,6 +386,9 @@ class TrafficGenerator(object):
             self._received = self._result.split(', ')[1].split('=')[1]
             self._sent = self._result.split(', ')[2].split('=')[1]
             self._loss = self._result.split(', ')[3].split('=')[1]
+            self._latency = []
+            self._latency.append(self._result.split(', ')[4].split('=')[1])
+            self._latency.append(self._result.split(', ')[5].split('=')[1])
 
     def stop_traffic_on_tg(self):
         """Stop all traffic on TG
@@ -380,17 +401,20 @@ class TrafficGenerator(object):
             self.trex_stl_stop_remote_exec(self._node)
 
     def send_traffic_on_tg(self, duration, rate, framesize,
-                           traffic_type, warmup_time=5, async_call=False):
+                           traffic_type, warmup_time=5, async_call=False,
+                           latency=True):
         """Send traffic from all configured interfaces on TG.
 
         :param duration: Duration of test traffic generation in seconds.
         :param rate: Offered load per interface (e.g. 1%, 3gbps, 4mpps, ...).
         :param framesize: Frame size (L2) in Bytes.
         :param traffic_type: Traffic profile.
+        :param latency: With latency measurement.
         :type duration: str
         :type rate: str
         :type framesize: str
         :type traffic_type: str
+        :type latency: bool
         :return: TG output.
         :rtype: str
         """
@@ -406,7 +430,7 @@ class TrafficGenerator(object):
             raise Exception('TG subtype not defined')
         elif node['subtype'] == NodeSubTypeTG.TREX:
             self.trex_stl_start_remote_exec(duration, rate, framesize,
-                                            traffic_type, async_call,
+                                            traffic_type, async_call, latency,
                                             warmup_time=warmup_time)
         else:
             raise NotImplementedError("TG subtype not supported")
diff --git a/resources/libraries/robot/performance.robot b/resources/libraries/robot/performance.robot
index 4dcd84e223..63295b7159 100644
--- a/resources/libraries/robot/performance.robot
+++ b/resources/libraries/robot/performance.robot
@@ -343,8 +343,9 @@
 | | Set Search Frame Size | ${framesize}
 | | Set Search Rate Type pps
 | | Linear Search | ${start_rate} | ${topology_type}
-| | ${rate_per_stream}= | Verify Search Result
+| | ${rate_per_stream} | ${latency}= | Verify Search Result
 | | Display result of NDR search | ${rate_per_stream} | ${framesize} | 2
+| | ...                          | ${latency}
 | | Traffic should pass with no loss | ${duration} | ${rate_per_stream}pps
 | | ...                              | ${framesize} | ${topology_type}
 | | ...                              | fail_on_loss=${False}
@@ -383,9 +384,10 @@
 | | Run Keyword If | '${loss_acceptance_type}' == 'percentage'
 | | ...            | Set Loss Acceptance Type Percentage
 | | Linear Search | ${start_rate} | ${topology_type}
-| | ${rate_per_stream}= | Verify Search Result
+| | ${rate_per_stream} | ${latency}= | Verify Search Result
 | | Display result of PDR search | ${rate_per_stream} | ${framesize} | 2
 | | ...                          | ${loss_acceptance} | ${loss_acceptance_type}
+| | ...                          | ${latency}
 | | Traffic should pass with partial loss | ${duration} | ${rate_per_stream}pps
 | | ...                                   | ${framesize} | ${topology_type}
 | | ...                                   | ${loss_acceptance}
@@ -421,8 +423,9 @@
 | | Set Search Rate Type pps
 | | Set Binary Convergence Threshold | ${threshold}
 | | Binary Search | ${binary_min} | ${binary_max} | ${topology_type}
-| | ${rate_per_stream}= | Verify Search Result
+| | ${rate_per_stream} | ${latency}= | Verify Search Result
 | | Display result of NDR search | ${rate_per_stream} | ${framesize} | 2
+| | ...                          | ${latency}
 | | Traffic should pass with no loss | ${duration} | ${rate_per_stream}pps
 | | ...                              | ${framesize} | ${topology_type}
 | | ...                              | fail_on_loss=${False}
@@ -463,9 +466,10 @@
 | | ...            | Set Loss Acceptance Type Percentage
 | | Set Binary Convergence Threshold | ${threshold}
 | | Binary Search | ${binary_min} | ${binary_max} | ${topology_type}
-| | ${rate_per_stream}= | Verify Search Result
+| | ${rate_per_stream} | ${latency}= | Verify Search Result
 | | Display result of PDR search | ${rate_per_stream} | ${framesize} | 2
 | | ...                          | ${loss_acceptance} | ${loss_acceptance_type}
+| | ...                          | ${latency}
 | | Traffic should pass with partial loss | ${duration} | ${rate_per_stream}pps
 | | ...                                   | ${framesize} | ${topology_type}
 | | ...                                   | ${loss_acceptance}
@@ -502,8 +506,9 @@
 | | Set Search Rate Type pps
 | | Set Binary Convergence Threshold | ${threshold}
 | | Combined Search | ${start_rate} | ${topology_type}
-| | ${rate_per_stream}= | Verify Search Result
+| | ${rate_per_stream} | ${latency}= | Verify Search Result
 | | Display result of NDR search | ${rate_per_stream} | ${framesize} | 2
+| | ...                          | ${latency}
 | | Traffic should pass with no loss | ${duration} | ${rate_per_stream}pps
 | | ...                              | ${framesize} | ${topology_type}
 | | ...                              | fail_on_loss=${False}
@@ -546,9 +551,10 @@
 | | ...            | Set Loss Acceptance Type Percentage
 | | Set Binary Convergence Threshold | ${threshold}
 | | Combined Search | ${start_rate} | ${topology_type}
-| | ${rate_per_stream}= | Verify Search Result
+| | ${rate_per_stream} | ${latency}= | Verify Search Result
 | | Display result of PDR search | ${rate_per_stream} | ${framesize} | 2
 | | ...                          | ${loss_acceptance} | ${loss_acceptance_type}
+| | ...                          | ${latency}
 | | Traffic should pass with partial loss | ${duration} | ${rate_per_stream}pps
 | | ...                                   | ${framesize} | ${topology_type}
 | | ...                                   | ${loss_acceptance}
@@ -563,14 +569,17 @@
 | | ... | - ${rate_per_stream} - Measured rate per stream [pps]. Type: string
 | | ... | - ${framesize} - L2 Frame Size [B]. Type: integer
 | | ... | - ${nr_streams} - Total number of streams. Type: integer
+| | ... | - ${latency} - Latency stats. Type: dictionary
 | | ...
 | | ... | *Return:*
 | | ... | - No value returned
 | | ...
 | | ... | *Example:*
 | | ...
-| | ... | \| Display result of NDR search \| 4400000 \| 64 \| 2
+| | ... | \| Display result of NDR search \| 4400000 \| 64 \| 2 \
+| | ... | \| (0, 10/10/10) \|
 | | [Arguments] | ${rate_per_stream} | ${framesize} | ${nr_streams}
+| | ...         | ${latency}
 | | ${rate_total}= | Evaluate | ${rate_per_stream}*${nr_streams}
 | | ${bandwidth_total}= | Evaluate | ${rate_total}*(${framesize}+20)*8/(10**9)
 | | Set Test Message | FINAL_RATE: ${rate_total} pps
@@ -578,6 +587,9 @@
 | | ...              | append=yes
 | | Set Test Message | ${\n}FINAL_BANDWIDTH: ${bandwidth_total} Gbps (untagged)
 | | ...              | append=yes
+| | :FOR | ${idx} | ${lat} | IN ENUMERATE | @{latency}
+| | | Set Test Message | ${\n}LATENCY_STREAM_${idx}: ${lat} usec (min/avg/max)
+| | ...                | append=yes
 
 | Display result of PDR search
 | | [Documentation] | Display result of PDR search in packet per seconds (total
@@ -589,6 +601,7 @@
 | | ... | - ${nr_streams} - Total number of streams. Type: integer
 | | ... | - ${loss_acceptance} - Accepted loss during search. Type: float
 | | ... | - ${loss_acceptance_type} - Percentage or frames. Type: string
+| | ... | - ${latency} - Latency stats. Type: dictionary
 | | ...
 | | ... | *Return:*
 | | ... | - No value returned
@@ -596,9 +609,9 @@
 | | ... | *Example:*
 | | ...
 | | ... | \| Display result of PDR search \| 4400000 \| 64 \| 2 \| 0.5 \
-| | ... | \| percentage
+| | ... | \| percentage \| (0, 10/10/10) \|
 | | [Arguments] | ${rate_per_stream} | ${framesize} | ${nr_streams}
-| | ...         | ${loss_acceptance} | ${loss_acceptance_type}
+| | ...         | ${loss_acceptance} | ${loss_acceptance_type} | ${latency}
 | | ${rate_total}= | Evaluate | ${rate_per_stream}*${nr_streams}
 | | ${bandwidth_total}= | Evaluate | ${rate_total}*(${framesize}+20)*8/(10**9)
 | | Set Test Message | FINAL_RATE: ${rate_total} pps
@@ -606,6 +619,9 @@
 | | ...              | append=yes
 | | Set Test Message | ${\n}FINAL_BANDWIDTH: ${bandwidth_total} Gbps (untagged)
 | | ...              | append=yes
+| | :FOR | ${idx} | ${lat} | IN ENUMERATE | @{latency}
+| | | Set Test Message | ${\n}LATENCY_STREAM_${idx}: ${lat} usec (min/avg/max)
+| | ...                | append=yes
 | | Set Test Message | ${\n}LOSS_ACCEPTANCE: ${loss_acceptance} ${loss_acceptance_type}
 | | ...              | append=yes
 
@@ -670,7 +686,8 @@
 | Clear and show runtime counters with running traffic
 | | [Arguments] | ${duration} | ${rate} | ${framesize} | ${topology_type}
 | | Send traffic on tg | -1 | ${rate} | ${framesize}
-| | ...                | ${topology_type} | warmup_time=0 | async_call=True
+| | ...                | ${topology_type} | warmup_time=0 | async_call=${True}
+| | ...                | latency=${False}
 | | Clear runtime counters on all DUTs
 | | Sleep | ${duration}
 | | Show runtime counters on all DUTs
diff --git a/resources/tools/t-rex/t-rex-stateless.py b/resources/tools/t-rex/t-rex-stateless.py
index 77895d7b68..7ff4c3713c 100755
--- a/resources/tools/t-rex/t-rex-stateless.py
+++ b/resources/tools/t-rex/t-rex-stateless.py
@@ -109,7 +109,7 @@ def create_packets(traffic_options, frame_size=64):
     :type traffic_options: list
     :type frame_size: int
     :return: Packet instances.
-    :rtype STLPktBuilder
+    :rtype: Tuple of STLPktBuilder
     """
 
     if frame_size < 64:
@@ -150,11 +150,15 @@ def create_packets(traffic_options, frame_size=64):
                      ], split_by_field="src")
 
     pkt_a = STLPktBuilder(pkt=base_pkt_a/generate_payload(
-        fsize_no_fcs-len(base_pkt_a)), vm=vm1)
+        max(0, fsize_no_fcs-len(base_pkt_a))), vm=vm1)
     pkt_b = STLPktBuilder(pkt=base_pkt_b/generate_payload(
-        fsize_no_fcs-len(base_pkt_b)), vm=vm2)
+        max(0, fsize_no_fcs-len(base_pkt_b))), vm=vm2)
+    lat_a = STLPktBuilder(pkt=base_pkt_a/generate_payload(
+        max(0, fsize_no_fcs-len(base_pkt_a))))
+    lat_b = STLPktBuilder(pkt=base_pkt_b/generate_payload(
+        max(0, fsize_no_fcs-len(base_pkt_b))))
 
-    return(pkt_a, pkt_b)
+    return(pkt_a, pkt_b, lat_a, lat_b)
 
 
 def create_packets_v6(traffic_options, frame_size=78):
@@ -165,7 +169,7 @@ def create_packets_v6(traffic_options, frame_size=78):
     :type traffic_options: List
     :type frame_size: int
     :return: Packet instances.
-    :rtype STLPktBuilder
+    :rtype: Tuple of STLPktBuilder
     """
 
     if frame_size < 78:
@@ -213,25 +217,36 @@ def create_packets_v6(traffic_options, frame_size=78):
         max(0, fsize_no_fcs-len(base_pkt_a))), vm=vm1)
     pkt_b = STLPktBuilder(pkt=base_pkt_b/generate_payload(
         max(0, fsize_no_fcs-len(base_pkt_b))), vm=vm2)
+    lat_a = STLPktBuilder(pkt=base_pkt_a/generate_payload(
+        max(0, fsize_no_fcs-len(base_pkt_a))))
+    lat_b = STLPktBuilder(pkt=base_pkt_b/generate_payload(
+        max(0, fsize_no_fcs-len(base_pkt_b))))
 
-    return(pkt_a, pkt_b)
+    return(pkt_a, pkt_b, lat_a, lat_b)
 
 
-def simple_burst(pkt_a, pkt_b, duration, rate, warmup_time, async_start):
+def simple_burst(pkt_a, pkt_b, pkt_lat_a, pkt_lat_b, duration, rate,
+                 warmup_time, async_start, latency):
     """Run the traffic with specific parameters.
 
     :param pkt_a: Base packet for stream 1.
     :param pkt_b: Base packet for stream 2.
+    :param pkt_lat_a: Base packet for latency stream 1.
+    :param pkt_lat_b: Base packet for latency stream 2.
     :param duration: Duration of traffic run in seconds (-1=infinite).
     :param rate: Rate of traffic run [percentage, pps, bps].
     :param warmup_time: Warm up duration.
-    :async_start: Start the traffic and exit
+    :param async_start: Start the traffic and exit.
+    :param latency: With latency stats.
     :type pkt_a: STLPktBuilder
     :type pkt_b: STLPktBuilder
+    :type pkt_lat_a: STLPktBuilder
+    :type pkt_lat_b: STLPktBuilder
     :type duration: int
     :type rate: string
     :type warmup_time: int
     :type async_start: bool
+    :type latency: bool
     :return: nothing
     """
 
@@ -242,37 +257,60 @@ def simple_burst(pkt_a, pkt_b, duration, rate, warmup_time, async_start):
     total_sent = 0
     lost_a = 0
     lost_b = 0
+    lat_a = 'NA'
+    lat_b = 'NA'
 
     try:
         # turn this off if too many logs
         #client.set_verbose("high")
 
-        # create two streams
-        stream1 = STLStream(packet=pkt_a,
-                            mode=STLTXCont(pps=100))
-
-        # second stream with a phase of 10ns (inter stream gap)
-        stream2 = STLStream(packet=pkt_b,
-                            isg=10.0,
-                            mode=STLTXCont(pps=100))
-
         # connect to server
         client.connect()
 
         # prepare our ports (my machine has 0 <--> 1 with static route)
         client.reset(ports=[0, 1])
 
-        # add both streams to ports
+        # create two traffic streams without latency stats
+        stream1 = STLStream(packet=pkt_a,
+                            mode=STLTXCont(pps=1000))
+        # second traffic stream with a phase of 10ns (inter stream gap)
+        stream2 = STLStream(packet=pkt_b,
+                            isg=10.0,
+                            mode=STLTXCont(pps=1000))
         client.add_streams(stream1, ports=[0])
         client.add_streams(stream2, ports=[1])
 
+        if latency:
+            # create two traffic streams with latency stats
+            lat_stream1 = STLStream(packet=pkt_lat_a,
+                                    flow_stats=STLFlowLatencyStats(pg_id=0),
+                                    mode=STLTXCont(pps=1000))
+            # second traffic stream with a phase of 10ns (inter stream gap)
+            lat_stream2 = STLStream(packet=pkt_lat_b,
+                                    isg=10.0,
+                                    flow_stats=STLFlowLatencyStats(pg_id=1),
+                                    mode=STLTXCont(pps=1000))
+            client.add_streams(lat_stream1, ports=[0])
+            client.add_streams(lat_stream2, ports=[1])
+
         #warmup phase
         if warmup_time > 0:
+            # clear the stats before injecting
             client.clear_stats()
+
+            # choose rate and start traffic
             client.start(ports=[0, 1], mult=rate, duration=warmup_time)
+
+            # block until done
             client.wait_on_traffic(ports=[0, 1], timeout=(warmup_time+30))
+
+            if client.get_warnings():
+                for warning in client.get_warnings():
+                    print_error(warning)
+
+            # read the stats after the test
             stats = client.get_stats()
-            print stats
+
             print "#####warmup statistics#####"
             print json.dumps(stats, indent=4,
                              separators=(',', ': '), sort_keys=True)
@@ -282,11 +320,8 @@ def simple_burst(pkt_a, pkt_b, duration, rate, warmup_time, async_start):
             print "\npackets lost from 0 --> 1:   {0} pkts".format(lost_a)
             print "packets lost from 1 --> 0:   {0} pkts".format(lost_b)
 
-
         # clear the stats before injecting
         client.clear_stats()
-        total_rcvd = 0
-        total_sent = 0
         lost_a = 0
         lost_b = 0
 
@@ -297,16 +332,29 @@ def simple_burst(pkt_a, pkt_b, duration, rate, warmup_time, async_start):
             # block until done
             client.wait_on_traffic(ports=[0, 1], timeout=(duration+30))
 
+            if client.get_warnings():
+                for warning in client.get_warnings():
+                    print_error(warning)
+
             # read the stats after the test
             stats = client.get_stats()
 
             print "#####statistics#####"
             print json.dumps(stats, indent=4,
                              separators=(',', ': '), sort_keys=True)
-
             lost_a = stats[0]["opackets"] - stats[1]["ipackets"]
             lost_b = stats[1]["opackets"] - stats[0]["ipackets"]
 
+            if latency:
+                lat_a = "/".join((\
+                    str(stats["latency"][0]["latency"]["total_min"]),\
+                    str(stats["latency"][0]["latency"]["average"]),\
+                    str(stats["latency"][0]["latency"]["total_max"])))
+                lat_b = "/".join((\
+                    str(stats["latency"][1]["latency"]["total_min"]),\
+                    str(stats["latency"][1]["latency"]["average"]),\
+                    str(stats["latency"][1]["latency"]["total_max"])))
+
             total_sent = stats[0]["opackets"] + stats[1]["opackets"]
             total_rcvd = stats[0]["ipackets"] + stats[1]["ipackets"]
 
@@ -322,8 +370,11 @@ def simple_burst(pkt_a, pkt_b, duration, rate, warmup_time, async_start):
             client.disconnect(stop_traffic=False, release_ports=True)
         else:
             client.disconnect()
-            print "rate={0}, totalReceived={1}, totalSent={2}, frameLoss={3}"\
-                .format(rate, total_rcvd, total_sent, lost_a+lost_b)
+            print "rate={0}, totalReceived={1}, totalSent={2}, "\
+                "frameLoss={3}, latencyStream0(usec)={4}, "\
+                "latencyStream1(usec)={5}".format(rate, total_rcvd,\
+                total_sent, lost_a+lost_b, lat_a, lat_b)
+
 
 def print_error(msg):
     """Print error message on stderr.
@@ -336,16 +387,12 @@ def print_error(msg):
     sys.stderr.write(msg+'\n')
 
 
-def main():
-    """Main function."""
+def parse_args():
+    """Parse arguments from cmd line.
 
-    _traffic_options = {}
-    #default L3 profile is IPv4
-    _use_ipv6 = False
-    #default warmup time is 5 seconds
-    _warmup_time = 5
-    #default behaviour of this script is sychronous
-    _async_call = False
+    :return: Parsed arguments.
+    :rtype ArgumentParser
+    """
 
     parser = argparse.ArgumentParser()
     parser.add_argument("-d", "--duration", required=True, type=int,
@@ -355,10 +402,16 @@ def main():
     parser.add_argument("-r", "--rate", required=True,
                         help="Traffic rate with included units (%, pps)")
     parser.add_argument("-6", "--use_IPv6", action="store_true",
+                        default=False,
                         help="Use IPv6 traffic profile instead of IPv4")
     parser.add_argument("--async", action="store_true",
+                        default=False,
                         help="Non-blocking call of the script")
+    parser.add_argument("--latency", action="store_true",
+                        default=False,
+                        help="Add latency stream")
     parser.add_argument("-w", "--warmup_time", type=int,
+                        default=5,
                         help="Traffic warmup time in seconds, 0 = disable")
 #    parser.add_argument("--p1_src_mac",
 #                        help="Port 1 source MAC address")
@@ -385,29 +438,40 @@ def main():
 #    parser.add_argument("--p2_dst_end_ip",
 #                        help="Port 2 destination end IP address")
 
-    args = parser.parse_args()
+    return parser.parse_args()
+
+
+def main():
+    """Main function."""
+
+    args = parse_args()
 
     _duration = args.duration
     _frame_size = args.frame_size
     _rate = args.rate
     _use_ipv6 = args.use_IPv6
     _async_call = args.async
+    _latency = args.latency
+    _warmup_time = args.warmup_time
 
-    if args.warmup_time is not None:
-        _warmup_time = args.warmup_time
-
+    _traffic_options = {}
     for attr in [a for a in dir(args) if a.startswith('p')]:
         if getattr(args, attr) is not None:
             _traffic_options[attr] = getattr(args, attr)
 
     if _use_ipv6:
-        pkt_a, pkt_b = create_packets_v6(_traffic_options,
-                                         frame_size=_frame_size)
+        # WARNING: Trex limitation to IPv4 only. IPv6 is not yet supported.
+        print_error('IPv6 latency is not supported yet. Running without lat.')
+        _latency = False
+
+        pkt_a, pkt_b, lat_a, lat_b = create_packets_v6(_traffic_options,
+                                                       frame_size=_frame_size)
     else:
-        pkt_a, pkt_b = create_packets(_traffic_options,
-                                      frame_size=_frame_size)
+        pkt_a, pkt_b, lat_a, lat_b = create_packets(_traffic_options,
+                                                    frame_size=_frame_size)
 
-    simple_burst(pkt_a, pkt_b, _duration, _rate, _warmup_time, _async_call)
+    simple_burst(pkt_a, pkt_b, lat_a, lat_b, _duration, _rate, _warmup_time,
+                 _async_call, _latency)
 
 if __name__ == "__main__":
     sys.exit(main())