aboutsummaryrefslogtreecommitdiffstats
path: root/yardstick/network_services/helpers/vpp_helpers/multiple_loss_ratio_search.py
blob: 582e3dc276700457527e8ba4d69654c11ce30695 (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
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
# Copyright (c) 2019 Viosoft Corporation
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# This is a modified copy of
# https://gerrit.fd.io/r/gitweb?p=csit.git;a=blob_plain;f=resources/libraries/python/MLRsearch/MultipleLossRatioSearch.py;hb=HEAD

import datetime
import logging
import math
import time

from yardstick.network_services.helpers.vpp_helpers.abstract_search_algorithm import \
    AbstractSearchAlgorithm
from yardstick.network_services.helpers.vpp_helpers.ndr_pdr_result import \
    NdrPdrResult
from yardstick.network_services.helpers.vpp_helpers.receive_rate_interval import \
    ReceiveRateInterval
from yardstick.network_services.helpers.vpp_helpers.receive_rate_measurement import \
    ReceiveRateMeasurement

LOGGING = logging.getLogger(__name__)


class MultipleLossRatioSearch(AbstractSearchAlgorithm):
    """Optimized binary search algorithm for finding NDR and PDR bounds.

    Traditional binary search algorithm needs initial interval
    (lower and upper bound), and returns final interval after bisecting
    (until some exit condition is met).
    The exit condition is usually related to the interval width,
    (upper bound value minus lower bound value).

    The optimized algorithm contains several improvements
    aimed to reduce overall search time.

    One improvement is searching for two intervals at once.
    The intervals are for NDR (No Drop Rate) and PDR (Partial Drop Rate).

    Next improvement is that the initial interval does not need to be valid.
    Imagine initial interval (10, 11) where 11 is smaller
    than the searched value.
    The algorithm will try (11, 13) interval next, and if 13 is still smaller,
    (13, 17) and so on, doubling width until the upper bound is valid.
    The part when interval expands is called external search,
    the part when interval is bisected is called internal search.

    Next improvement is that trial measurements at small trial duration
    can be used to find a reasonable interval for full trial duration search.
    This results in more trials performed, but smaller overall duration
    in general.

    Next improvement is bisecting in logarithmic quantities,
    so that exit criteria can be independent of measurement units.

    Next improvement is basing the initial interval on receive rates.

    Final improvement is exiting early if the minimal value
    is not a valid lower bound.

    The complete search consist of several phases,
    each phase performing several trial measurements.
    Initial phase creates initial interval based on receive rates
    at maximum rate and at maximum receive rate (MRR).
    Final phase and preceding intermediate phases are performing
    external and internal search steps,
    each resulting interval is the starting point for the next phase.
    The resulting interval of final phase is the result of the whole algorithm.

    Each non-initial phase uses its own trial duration and width goal.
    Any non-initial phase stops searching (for NDR or PDR independently)
    when minimum is not a valid lower bound (at current duration),
    or all of the following is true:
    Both bounds are valid, bound bounds are measured at the current phase
    trial duration, interval width is less than the width goal
    for current phase.

    TODO: Review and update this docstring according to rst docs.
    TODO: Support configurable number of Packet Loss Ratios.
    """

    class ProgressState(object):
        """Structure containing data to be passed around in recursion."""

        def __init__(
                self, result, phases, duration, width_goal, packet_loss_ratio,
                minimum_transmit_rate, maximum_transmit_rate):
            """Convert and store the argument values.

            :param result: Current measured NDR and PDR intervals.
            :param phases: How many intermediate phases to perform
                before the current one.
            :param duration: Trial duration to use in the current phase [s].
            :param width_goal: The goal relative width for the curreent phase.
            :param packet_loss_ratio: PDR fraction for the current search.
            :param minimum_transmit_rate: Minimum target transmit rate
                for the current search [pps].
            :param maximum_transmit_rate: Maximum target transmit rate
                for the current search [pps].
            :type result: NdrPdrResult.NdrPdrResult
            :type phases: int
            :type duration: float
            :type width_goal: float
            :type packet_loss_ratio: float
            :type minimum_transmit_rate: float
            :type maximum_transmit_rate: float
            """
            self.result = result
            self.phases = int(phases)
            self.duration = float(duration)
            self.width_goal = float(width_goal)
            self.packet_loss_ratio = float(packet_loss_ratio)
            self.minimum_transmit_rate = float(minimum_transmit_rate)
            self.maximum_transmit_rate = float(maximum_transmit_rate)

    def __init__(self, measurer, latency=False, pkt_size=64,
                 final_relative_width=0.005,
                 final_trial_duration=30.0, initial_trial_duration=1.0,
                 number_of_intermediate_phases=2, timeout=600.0, doublings=1):
        """Store the measurer object and additional arguments.

        :param measurer: Rate provider to use by this search object.
        :param final_relative_width: Final lower bound transmit rate
            cannot be more distant that this multiple of upper bound [1].
        :param final_trial_duration: Trial duration for the final phase [s].
        :param initial_trial_duration: Trial duration for the initial phase
            and also for the first intermediate phase [s].
        :param number_of_intermediate_phases: Number of intermediate phases
            to perform before the final phase [1].
        :param timeout: The search will fail itself when not finished
            before this overall time [s].
        :param doublings: How many doublings to do in external search step.
            Default 1 is suitable for fairly stable tests,
            less stable tests might get better overal duration with 2 or more.
        :type measurer: AbstractMeasurer.AbstractMeasurer
        :type final_relative_width: float
        :type final_trial_duration: float
        :type initial_trial_duration: int
        :type number_of_intermediate_phases: int
        :type timeout: float
        :type doublings: int
        """
        super(MultipleLossRatioSearch, self).__init__(measurer)
        self.latency = latency
        self.pkt_size = int(pkt_size)
        self.final_trial_duration = float(final_trial_duration)
        self.final_relative_width = float(final_relative_width)
        self.number_of_intermediate_phases = int(number_of_intermediate_phases)
        self.initial_trial_duration = float(initial_trial_duration)
        self.timeout = float(timeout)
        self.doublings = int(doublings)

        self.queue = None
        self.port_pg_id = None
        self.ports = []
        self.test_data = {}
        self.profiles = {}

    @staticmethod
    def double_relative_width(relative_width):
        """Return relative width corresponding to double logarithmic width.

        :param relative_width: The base relative width to double.
        :type relative_width: float
        :returns: The relative width of double logarithmic size.
        :rtype: float
        """
        return 1.999 * relative_width - relative_width * relative_width
        # The number should be 2.0, but we want to avoid rounding errors,
        # and ensure half of double is not larger than the original value.

    @staticmethod
    def double_step_down(relative_width, current_bound):
        """Return rate of double logarithmic width below.

        :param relative_width: The base relative width to double.
        :param current_bound: The current target transmit rate to move [pps].
        :type relative_width: float
        :type current_bound: float
        :returns: Transmit rate smaller by logarithmically double width [pps].
        :rtype: float
        """
        return current_bound * (
                1.0 - MultipleLossRatioSearch.double_relative_width(
            relative_width))

    @staticmethod
    def expand_down(relative_width, doublings, current_bound):
        """Return rate of expanded logarithmic width below.

        :param relative_width: The base relative width to double.
        :param doublings: How many doublings to do for expansion.
        :param current_bound: The current target transmit rate to move [pps].
        :type relative_width: float
        :type doublings: int
        :type current_bound: float
        :returns: Transmit rate smaller by logarithmically double width [pps].
        :rtype: float
        """
        for _ in range(doublings):
            relative_width = MultipleLossRatioSearch.double_relative_width(
                relative_width)
        return current_bound * (1.0 - relative_width)

    @staticmethod
    def double_step_up(relative_width, current_bound):
        """Return rate of double logarithmic width above.

        :param relative_width: The base relative width to double.
        :param current_bound: The current target transmit rate to move [pps].
        :type relative_width: float
        :type current_bound: float
        :returns: Transmit rate larger by logarithmically double width [pps].
        :rtype: float
        """
        return current_bound / (
                1.0 - MultipleLossRatioSearch.double_relative_width(
            relative_width))

    @staticmethod
    def expand_up(relative_width, doublings, current_bound):
        """Return rate of expanded logarithmic width above.

        :param relative_width: The base relative width to double.
        :param doublings: How many doublings to do for expansion.
        :param current_bound: The current target transmit rate to move [pps].
        :type relative_width: float
        :type doublings: int
        :type current_bound: float
        :returns: Transmit rate smaller by logarithmically double width [pps].
        :rtype: float
        """
        for _ in range(doublings):
            relative_width = MultipleLossRatioSearch.double_relative_width(
                relative_width)
        return current_bound / (1.0 - relative_width)

    @staticmethod
    def half_relative_width(relative_width):
        """Return relative width corresponding to half logarithmic width.

        :param relative_width: The base relative width to halve.
        :type relative_width: float
        :returns: The relative width of half logarithmic size.
        :rtype: float
        """
        return 1.0 - math.sqrt(1.0 - relative_width)

    @staticmethod
    def half_step_up(relative_width, current_bound):
        """Return rate of half logarithmic width above.

        :param relative_width: The base relative width to halve.
        :param current_bound: The current target transmit rate to move [pps].
        :type relative_width: float
        :type current_bound: float
        :returns: Transmit rate larger by logarithmically half width [pps].
        :rtype: float
        """
        return current_bound / (
                1.0 - MultipleLossRatioSearch.half_relative_width(
            relative_width))

    def init_generator(self, ports, port_pg_id, profiles, test_data, queue):
        self.ports = ports
        self.port_pg_id = port_pg_id
        self.profiles = profiles
        self.test_data = test_data
        self.queue = queue
        self.queue.cancel_join_thread()

    def collect_kpi(self, stats, test_value):
        samples = self.measurer.generate_samples(stats, self.ports,
                                                 self.port_pg_id, self.latency)
        samples.update(self.test_data)
        LOGGING.info("Collect TG KPIs %s %s %s", datetime.datetime.now(),
                     test_value, samples)
        self.queue.put(samples)

    def narrow_down_ndr_and_pdr(
            self, minimum_transmit_rate, maximum_transmit_rate,
            packet_loss_ratio):
        """Perform initial phase, create state object, proceed with next phases.

        :param minimum_transmit_rate: Minimal target transmit rate [pps].
        :param maximum_transmit_rate: Maximal target transmit rate [pps].
        :param packet_loss_ratio: Fraction of packets lost, for PDR [1].
        :type minimum_transmit_rate: float
        :type maximum_transmit_rate: float
        :type packet_loss_ratio: float
        :returns: Structure containing narrowed down intervals
            and their measurements.
        :rtype: NdrPdrResult.NdrPdrResult
        :raises RuntimeError: If total duration is larger than timeout.
        """
        minimum_transmit_rate = float(minimum_transmit_rate)
        maximum_transmit_rate = float(maximum_transmit_rate)
        packet_loss_ratio = float(packet_loss_ratio)
        line_measurement = self.measure(
            self.initial_trial_duration, maximum_transmit_rate, self.latency)
        initial_width_goal = self.final_relative_width
        for _ in range(self.number_of_intermediate_phases):
            initial_width_goal = self.double_relative_width(initial_width_goal)
        max_lo = maximum_transmit_rate * (1.0 - initial_width_goal)
        mrr = max(
            minimum_transmit_rate,
            min(max_lo, line_measurement.receive_rate))
        mrr_measurement = self.measure(
            self.initial_trial_duration, mrr, self.latency)
        # Attempt to get narrower width.
        if mrr_measurement.loss_fraction > 0.0:
            max2_lo = mrr * (1.0 - initial_width_goal)
            mrr2 = min(max2_lo, mrr_measurement.receive_rate)
        else:
            mrr2 = mrr / (1.0 - initial_width_goal)
        if mrr2 > minimum_transmit_rate and mrr2 < maximum_transmit_rate:
            line_measurement = mrr_measurement
            mrr_measurement = self.measure(
                self.initial_trial_duration, mrr2, self.latency)
            if mrr2 > mrr:
                buf = line_measurement
                line_measurement = mrr_measurement
                mrr_measurement = buf
        starting_interval = ReceiveRateInterval(
            mrr_measurement, line_measurement)
        starting_result = NdrPdrResult(starting_interval, starting_interval)
        state = self.ProgressState(
            starting_result, self.number_of_intermediate_phases,
            self.final_trial_duration, self.final_relative_width,
            packet_loss_ratio, minimum_transmit_rate, maximum_transmit_rate)
        state = self.ndrpdr(state)
        result = state.result
        # theor_max_thruput = 0
        result_samples = {}

        MultipleLossRatioSearch.display_single_bound(result_samples,
                'NDR_LOWER', result.ndr_interval.measured_low.transmit_rate,
                self.pkt_size, result.ndr_interval.measured_low.latency)
        MultipleLossRatioSearch.display_single_bound(result_samples,
                'NDR_UPPER', result.ndr_interval.measured_high.transmit_rate,
                self.pkt_size)
        MultipleLossRatioSearch.display_single_bound(result_samples,
                'PDR_LOWER', result.pdr_interval.measured_low.transmit_rate,
                self.pkt_size, result.pdr_interval.measured_low.latency)
        MultipleLossRatioSearch.display_single_bound(result_samples,
                'PDR_UPPER', result.pdr_interval.measured_high.transmit_rate,
                self.pkt_size)
        pdr_msg = self.check_ndrpdr_interval_validity(result_samples, "PDR",
                                                      result.pdr_interval,
                                                      packet_loss_ratio)
        ndr_msg = self.check_ndrpdr_interval_validity(result_samples, "NDR",
                                                      result.ndr_interval)
        self.queue.put(result_samples)

        LOGGING.debug("result_samples: %s", result_samples)
        LOGGING.info(pdr_msg)
        LOGGING.info(ndr_msg)

        self.perform_additional_measurements_based_on_ndrpdr_result(result)

        return result_samples

    def _measure_and_update_state(self, state, transmit_rate):
        """Perform trial measurement, update bounds, return new state.

        :param state: State before this measurement.
        :param transmit_rate: Target transmit rate for this measurement [pps].
        :type state: ProgressState
        :type transmit_rate: float
        :returns: State after the measurement.
        :rtype: ProgressState
        """
        # TODO: Implement https://stackoverflow.com/a/24683360
        # to avoid the string manipulation if log verbosity is too low.
        LOGGING.info("result before update: %s", state.result)
        LOGGING.debug(
            "relative widths in goals: %s", state.result.width_in_goals(
                self.final_relative_width))
        measurement = self.measure(state.duration, transmit_rate, self.latency)
        ndr_interval = self._new_interval(
            state.result.ndr_interval, measurement, 0.0)
        pdr_interval = self._new_interval(
            state.result.pdr_interval, measurement, state.packet_loss_ratio)
        state.result = NdrPdrResult(ndr_interval, pdr_interval)
        return state

    @staticmethod
    def _new_interval(old_interval, measurement, packet_loss_ratio):
        """Return new interval with bounds updated according to the measurement.

        :param old_interval: The current interval before the measurement.
        :param measurement: The new meaqsurement to take into account.
        :param packet_loss_ratio: Fraction for PDR (or zero for NDR).
        :type old_interval: ReceiveRateInterval.ReceiveRateInterval
        :type measurement: ReceiveRateMeasurement.ReceiveRateMeasurement
        :type packet_loss_ratio: float
        :returns: The updated interval.
        :rtype: ReceiveRateInterval.ReceiveRateInterval
        """
        old_lo, old_hi = old_interval.measured_low, old_interval.measured_high
        # Priority zero: direct replace if the target Tr is the same.
        if measurement.target_tr in (old_lo.target_tr, old_hi.target_tr):
            if measurement.target_tr == old_lo.target_tr:
                return ReceiveRateInterval(measurement, old_hi)
            else:
                return ReceiveRateInterval(old_lo, measurement)
        # Priority one: invalid lower bound allows only one type of update.
        if old_lo.loss_fraction > packet_loss_ratio:
            # We can only expand down, old bound becomes valid upper one.
            if measurement.target_tr < old_lo.target_tr:
                return ReceiveRateInterval(measurement, old_lo)
            else:
                return old_interval
        # Lower bound is now valid.
        # Next priorities depend on target Tr.
        if measurement.target_tr < old_lo.target_tr:
            # Lower external measurement, relevant only
            # if the new measurement has high loss rate.
            if measurement.loss_fraction > packet_loss_ratio:
                # Returning the broader interval as old_lo
                # would be invalid upper bound.
                return ReceiveRateInterval(measurement, old_hi)
        elif measurement.target_tr > old_hi.target_tr:
            # Upper external measurement, only relevant for invalid upper bound.
            if old_hi.loss_fraction <= packet_loss_ratio:
                # Old upper bound becomes valid new lower bound.
                return ReceiveRateInterval(old_hi, measurement)
        else:
            # Internal measurement, replaced boundary
            # depends on measured loss fraction.
            if measurement.loss_fraction > packet_loss_ratio:
                # We have found a narrow valid interval,
                # regardless of whether old upper bound was valid.
                return ReceiveRateInterval(old_lo, measurement)
            else:
                # In ideal world, we would not want to shrink interval
                # if upper bound is not valid.
                # In the real world, we want to shrink it for
                # "invalid upper bound at maximal rate" case.
                return ReceiveRateInterval(measurement, old_hi)
        # Fallback, the interval is unchanged by the measurement.
        return old_interval

    def ndrpdr(self, state):
        """Pefrom trials for this phase. Return the new state when done.

        :param state: State before this phase.
        :type state: ProgressState
        :returns: The updated state.
        :rtype: ProgressState
        :raises RuntimeError: If total duration is larger than timeout.
        """
        start_time = time.time()
        if state.phases > 0:
            # We need to finish preceding intermediate phases first.
            saved_phases = state.phases
            state.phases -= 1
            # Preceding phases have shorter duration.
            saved_duration = state.duration
            duration_multiplier = state.duration / self.initial_trial_duration
            phase_exponent = float(state.phases) / saved_phases
            state.duration = self.initial_trial_duration * math.pow(
                duration_multiplier, phase_exponent)
            # Shorter durations do not need that narrow widths.
            saved_width = state.width_goal
            state.width_goal = self.double_relative_width(state.width_goal)
            # Recurse.
            state = self.ndrpdr(state)
            # Restore the state for current phase.
            state.duration = saved_duration
            state.width_goal = saved_width
            state.phases = saved_phases  # Not needed, but just in case.
        LOGGING.info(
            "starting iterations with duration %s and relative width goal %s",
            state.duration, state.width_goal)
        while 1:
            if time.time() > start_time + self.timeout:
                raise RuntimeError("Optimized search takes too long.")
            # Order of priorities: invalid bounds (nl, pl, nh, ph),
            # then narrowing relative Tr widths.
            # Durations are not priorities yet,
            # they will settle on their own hopefully.
            ndr_lo = state.result.ndr_interval.measured_low
            ndr_hi = state.result.ndr_interval.measured_high
            pdr_lo = state.result.pdr_interval.measured_low
            pdr_hi = state.result.pdr_interval.measured_high
            ndr_rel_width = max(
                state.width_goal, state.result.ndr_interval.rel_tr_width)
            pdr_rel_width = max(
                state.width_goal, state.result.pdr_interval.rel_tr_width)
            # If we are hitting maximal or minimal rate, we cannot shift,
            # but we can re-measure.
            if ndr_lo.loss_fraction > 0.0:
                if ndr_lo.target_tr > state.minimum_transmit_rate:
                    new_tr = max(
                        state.minimum_transmit_rate,
                        self.expand_down(
                            ndr_rel_width, self.doublings, ndr_lo.target_tr))
                    LOGGING.info("ndr lo external %s", new_tr)
                    state = self._measure_and_update_state(state, new_tr)
                    continue
                elif ndr_lo.duration < state.duration:
                    LOGGING.info("ndr lo minimal re-measure")
                    state = self._measure_and_update_state(
                        state, state.minimum_transmit_rate)
                    continue
            if pdr_lo.loss_fraction > state.packet_loss_ratio:
                if pdr_lo.target_tr > state.minimum_transmit_rate:
                    new_tr = max(
                        state.minimum_transmit_rate,
                        self.expand_down(
                            pdr_rel_width, self.doublings, pdr_lo.target_tr))
                    LOGGING.info("pdr lo external %s", new_tr)
                    state = self._measure_and_update_state(state, new_tr)
                    continue
                elif pdr_lo.duration < state.duration:
                    LOGGING.info("pdr lo minimal re-measure")
                    state = self._measure_and_update_state(
                        state, state.minimum_transmit_rate)
                    continue
            if ndr_hi.loss_fraction <= 0.0:
                if ndr_hi.target_tr < state.maximum_transmit_rate:
                    new_tr = min(
                        state.maximum_transmit_rate,
                        self.expand_up(
                            ndr_rel_width, self.doublings, ndr_hi.target_tr))
                    LOGGING.info("ndr hi external %s", new_tr)
                    state = self._measure_and_update_state(state, new_tr)
                    continue
                elif ndr_hi.duration < state.duration:
                    LOGGING.info("ndr hi maximal re-measure")
                    state = self._measure_and_update_state(
                        state, state.maximum_transmit_rate)
                    continue
            if pdr_hi.loss_fraction <= state.packet_loss_ratio:
                if pdr_hi.target_tr < state.maximum_transmit_rate:
                    new_tr = min(
                        state.maximum_transmit_rate,
                        self.expand_up(
                            pdr_rel_width, self.doublings, pdr_hi.target_tr))
                    LOGGING.info("pdr hi external %s", new_tr)
                    state = self._measure_and_update_state(state, new_tr)
                    continue
                elif pdr_hi.duration < state.duration:
                    LOGGING.info("ndr hi maximal re-measure")
                    state = self._measure_and_update_state(
                        state, state.maximum_transmit_rate)
                    continue
            # If we are hitting maximum_transmit_rate,
            # it is still worth narrowing width,
            # hoping large enough loss fraction will happen.
            # But if we are hitting the minimal rate (at current duration),
            # no additional measurement will help with that,
            # so we can stop narrowing in this phase.
            if (ndr_lo.target_tr <= state.minimum_transmit_rate
                    and ndr_lo.loss_fraction > 0.0):
                ndr_rel_width = 0.0
            if (pdr_lo.target_tr <= state.minimum_transmit_rate
                    and pdr_lo.loss_fraction > state.packet_loss_ratio):
                pdr_rel_width = 0.0
            if ndr_rel_width > state.width_goal:
                # We have to narrow NDR width first, as NDR internal search
                # can invalidate PDR (but not vice versa).
                new_tr = self.half_step_up(ndr_rel_width, ndr_lo.target_tr)
                LOGGING.info("Bisecting for NDR at %s", new_tr)
                state = self._measure_and_update_state(state, new_tr)
                continue
            if pdr_rel_width > state.width_goal:
                # PDR iternal search.
                new_tr = self.half_step_up(pdr_rel_width, pdr_lo.target_tr)
                LOGGING.info("Bisecting for PDR at %s", new_tr)
                state = self._measure_and_update_state(state, new_tr)
                continue
            # We do not need to improve width, but there still might be
            # some measurements with smaller duration.
            # We need to re-measure with full duration, possibly
            # creating invalid bounds to resolve (thus broadening width).
            if ndr_lo.duration < state.duration:
                LOGGING.info("re-measuring NDR lower bound")
                state = self._measure_and_update_state(state, ndr_lo.target_tr)
                continue
            if pdr_lo.duration < state.duration:
                LOGGING.info("re-measuring PDR lower bound")
                state = self._measure_and_update_state(state, pdr_lo.target_tr)
                continue
            # Except when lower bounds have high loss fraction, in that case
            # we do not need to re-measure _upper_ bounds.
            if ndr_hi.duration < state.duration and ndr_rel_width > 0.0:
                LOGGING.info("re-measuring NDR upper bound")
                state = self._measure_and_update_state(state, ndr_hi.target_tr)
                continue
            if pdr_hi.duration < state.duration and pdr_rel_width > 0.0:
                LOGGING.info("re-measuring PDR upper bound")
                state = self._measure_and_update_state(state, pdr_hi.target_tr)
                continue
            # Widths are narrow (or lower bound minimal), bound measurements
            # are long enough, we can return.
            LOGGING.info("phase done")
            break
        return state

    def measure(self, duration, transmit_rate, latency):
        duration = float(duration)
        transmit_rate = float(transmit_rate)
        # Trex needs target Tr per stream, but reports aggregate Tx and Dx.
        unit_rate = str(transmit_rate / 2.0) + "pps"
        stats = self.measurer.send_traffic_on_tg(self.ports, self.port_pg_id,
                                                 duration, unit_rate,
                                                 latency=latency)
        self.measurer.client.reset(ports=self.ports)
        self.measurer.client.clear_stats(ports=self.ports)
        self.measurer.client.remove_all_streams(ports=self.ports)
        for port, profile in self.profiles.items():
            self.measurer.client.add_streams(profile, ports=[port])
        self.collect_kpi(stats, unit_rate)
        transmit_count = int(self.measurer.sent)
        loss_count = int(self.measurer.loss)
        measurement = ReceiveRateMeasurement(
            duration, transmit_rate, transmit_count, loss_count)
        measurement.latency = self.measurer.latency
        return measurement

    def perform_additional_measurements_based_on_ndrpdr_result(self, result):
        duration = 5.0
        rate = "{}{}".format(result.ndr_interval.measured_low.target_tr / 2.0,
                             'pps')
        for _ in range(0, 1):
            stats = self.measurer.send_traffic_on_tg(self.ports,
                                                     self.port_pg_id, duration,
                                                     rate)
            self.collect_kpi(stats, rate)
            LOGGING.info('Traffic loss occurred: %s', self.measurer.loss)

    @staticmethod
    def display_single_bound(result_samples, result_type, rate_total, pkt_size,
                             latency=None):
        bandwidth_total = float(rate_total) * (pkt_size + 20) * 8 / (10 ** 9)

        result_samples["Result_{}".format(result_type)] = {
            "rate_total_pps": float(rate_total),
            "bandwidth_total_Gbps": float(bandwidth_total),
        }

        if latency:
            for item in latency:
                if latency.index(item) == 0:
                    name = "Result_{}_{}".format("stream0", result_type)
                else:
                    name = "Result_{}_{}".format("stream1", result_type)
                lat_min, lat_avg, lat_max = item.split('/')
                result_samples[name] = {
                    "min_latency": float(lat_min),
                    "avg_latency": float(lat_avg),
                    "max_latency": float(lat_max),
                }

    @staticmethod
    def check_ndrpdr_interval_validity(result_samples, result_type, interval,
                                       packet_loss_ratio=0.0):
        lower_bound = interval.measured_low
        lower_bound_lf = lower_bound.loss_fraction

        result_samples["Result_{}_packets_lost".format(result_type)] = {
            "packet_loss_ratio": float(lower_bound_lf),
            "packets_lost": float(lower_bound.loss_count),
        }

        if lower_bound_lf <= packet_loss_ratio:
            return "Minimal rate loss fraction {} reach target {}".format(
                lower_bound_lf, packet_loss_ratio)
        else:
            message = "Minimal rate loss fraction {} does not reach target {}".format(
                lower_bound_lf, packet_loss_ratio)
            if lower_bound_lf >= 1.0:
                return '{}\nZero packets forwarded!'.format(message)
            else:
                return '{}\n{} packets lost.'.format(message,
                                                     lower_bound.loss_count)