8 The objective of the OPNFV project titled
9 **“Characterize vSwitch Performance for Telco NFV Use Cases”**, is to
10 evaluate a virtual switch to identify its suitability for a Telco
11 Network Function Virtualization (NFV) environment. The intention of this
12 Level Test Design (LTD) document is to specify the set of tests to carry
13 out in order to objectively measure the current characteristics of a
14 virtual switch in the Network Function Virtualization Infrastructure
15 (NFVI) as well as the test pass criteria. The detailed test cases will
16 be defined in `Section 2 <#DetailsOfTheLevelTestDesign>`__, preceded by
17 the `Document identifier <#DocId>`__ and the `Scope <#Scope>`__.
19 This document is currently in draft form.
24 =========================
26 The document id will be used to uniquely
27 identify versions of the LTD. The format for the document id will be:
28 OPNFV\_vswitchperf\_LTD\_ver\_NUM\_MONTH\_YEAR\_STATUS, where by the
29 status is one of: draft, reviewed, corrected or final. The document id
30 for this version of the LTD is:
31 OPNFV\_vswitchperf\_LTD\_ver\_1.6\_Jan\_15\_DRAFT.
38 The main purpose of this project is to specify a suite of
39 performance tests in order to objectively measure the current packet
40 transfer characteristics of a virtual switch in the NFVI. The intent of
41 the project is to facilitate testing of any virtual switch. Thus, a
42 generic suite of tests shall be developed, with no hard dependencies to
43 a single implementation. In addition, the test case suite shall be
44 architecture independent.
46 The test cases developed in this project shall not form part of a
47 separate test framework, all of these tests may be inserted into the
48 Continuous Integration Test Framework and/or the Platform Functionality
49 Test Framework - if a vSwitch becomes a standard component of an OPNFV
57 * `RFC 1242 Benchmarking Terminology for Network Interconnection
58 Devices <http://www.ietf.org/rfc/rfc1242.txt>`__
59 * `RFC 2544 Benchmarking Methodology for Network Interconnect
60 Devices <http://www.ietf.org/rfc/rfc2544.txt>`__
61 * `RFC 2285 Benchmarking Terminology for LAN Switching
62 Devices <http://www.ietf.org/rfc/rfc2285.txt>`__
63 * `RFC 2889 Benchmarking Methodology for LAN Switching
64 Devices <http://www.ietf.org/rfc/rfc2889.txt>`__
65 * `RFC 3918 Methodology for IP Multicast
66 Benchmarking <http://www.ietf.org/rfc/rfc3918.txt>`__
67 * `RFC 4737 Packet Reordering
68 Metrics <http://www.ietf.org/rfc/rfc4737.txt>`__
69 * `RFC 5481 Packet Delay Variation Applicability
70 Statement <http://www.ietf.org/rfc/rfc5481.txt>`__
71 * `RFC 6201 Device Reset
72 Characterization <http://tools.ietf.org/html/rfc6201>`__
76 ===================================
77 Details of the Level Test Design
78 ===================================
80 This section describes the features to be tested (
81 :ref:_FeaturesToBeTested), the test approach (:ref:_Approach);
82 it also identifies the sets of test cases or scenarios (
83 :ref:_TestIdentification) along with the pass/fail criteria and
84 the test deliverables.
88 .. _FeaturesToBeTested:
91 ==========================
93 Characterizing virtual switches (i.e. Device Under Test (DUT) in this document)
94 includes measuring the following performance metrics:
96 - **Throughput** as defined by `RFC1242
97 <https://www.rfc-editor.org/rfc/rfc1242.txt>`__: The maximum rate at which
98 **none** of the offered frames are dropped by the DUT. The maximum frame
99 rate and bit rate that can be transmitted by the DUT without any error
100 should be recorded. Note there is an equivalent bit rate and a specific
101 layer at which the payloads contribute to the bits. Errors and
102 improperly formed frames or packets are dropped.
103 - **Packet delay** introduced by the DUT and its cumulative effect on
104 E2E networks. Frame delay can be measured equivalently.
105 - **Packet delay variation**: measured from the perspective of the
106 VNF/application. Packet delay variation is sometimes called "jitter".
107 However, we will avoid the term "jitter" as the term holds different
108 meaning to different groups of people. In this document we will
109 simply use the term packet delay variation. The preferred form for this
110 metric is the PDV form of delay variation defined in `RFC5481
111 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__. The most relevant
112 measurement of PDV considers the delay variation of a single user flow,
113 as this will be relevant to the size of end-system buffers to compensate
114 for delay variation. The measurement system's ability to store the
115 delays of individual packets in the flow of interest is a key factor
116 that determines the specific measurement method. At the outset, it is
117 ideal to view the complete PDV distribution. Systems that can capture
118 and store packets and their delays have the freedom to calculate the
119 reference minimum delay and to determine various quantiles of the PDV
120 distribution accurately (in post-measurement processing routines).
121 Systems without storage must apply algorithms to calculate delay and
122 statistical measurements on the fly. For example, a system may store
123 temporary estimates of the mimimum delay and the set of (100) packets
124 with the longest delays during measurement (to calculate a high quantile,
125 and update these sets with new values periodically.
126 In some cases, a limited number of delay histogram bins will be
127 available, and the bin limits will need to be set using results from
128 repeated experiments. See section 8 of `RFC5481
129 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__.
130 - **Packet loss** (within a configured waiting time at the receiver): All
131 packets sent to the DUT should be accounted for.
132 - **Burst behaviour**: measures the ability of the DUT to buffer packets.
133 - **Packet re-ordering**: measures the ability of the device under test to
134 maintain sending order throughout transfer to the destination.
135 - **Packet correctness**: packets or Frames must be well-formed, in that
136 they include all required fields, conform to length requirements, pass
137 integrity checks, etc.
138 - **Availability and capacity** of the DUT i.e. when the DUT is fully “up”
139 and connected, following measurements should be captured for
140 DUT without any network packet load:
142 - Includes average power consumption of the CPUs (in various power states) and
143 system over specified period of time. Time period should not be less
145 - Includes average per core CPU utilization over specified period of time.
146 Time period should not be less than 60 seconds.
147 - Includes the number of NIC interfaces supported.
148 - Includes headroom of VM workload processing cores (i.e. available
158 In order to determine the packet transfer characteristics of a virtual
159 switch, the tests will be broken down into the following categories:
164 ----------------------
165 - **Throughput Tests** to measure the maximum forwarding rate (in
166 frames per second or fps) and bit rate (in Mbps) for a constant load
167 (as defined by `RFC1242 <https://www.rfc-editor.org/rfc/rfc1242.txt>`__)
168 without traffic loss.
169 - **Packet and Frame Delay Tests** to measure average, min and max
170 packet and frame delay for constant loads.
171 - **Stream Performance Tests** (TCP, UDP) to measure bulk data transfer
172 performance, i.e. how fast systems can send and receive data through
174 - **Request/Response Performance** Tests (TCP, UDP) the measure the
175 transaction rate through the virtual switch.
176 - **Packet Delay Tests** to understand latency distribution for
177 different packet sizes and over an extended test run to uncover
179 - **Scalability Tests** to understand how the virtual switch performs
180 as the number of flows, active ports, complexity of the forwarding
181 logic's configuration... it has to deal with increases.
182 - **Control Path and Datapath Coupling** Tests, to understand how
183 closely coupled the datapath and the control path are as well as the
184 effect of this coupling on the performance of the DUT.
185 - **CPU and Memory Consumption Tests** to understand the virtual
186 switch’s footprint on the system, this includes:
188 * CPU core utilization.
189 * CPU cache utilization.
191 * System bus (QPI, PCI, ..) utilization.
192 * Memory lanes utilization.
193 * CPU cycles consumed per packet.
194 * Time To Establish Flows Tests.
196 - **Noisy Neighbour Tests**, to understand the effects of resource
197 sharing on the performance of a virtual switch.
199 **Note:** some of the tests above can be conducted simultaneously where
200 the combined results would be insightful, for example Packet/Frame Delay
206 --------------------------
207 The following represents possible deployment test scenarios which can
208 help to determine the performance of both the virtual switch and the
209 datapaths to physical ports (to NICs) and to logical ports (to VNFs):
213 Physical port → vSwitch → physical port
214 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
215 .. code-block:: console
218 +--------------------------------------------------+ |
219 | +--------------------+ | |
222 | +--------------+ +--------------+ | |
223 | | phy port | vSwitch | phy port | | |
224 +---+--------------+------------+--------------+---+ _|
228 +--------------------------------------------------+
230 | traffic generator |
232 +--------------------------------------------------+
236 Physical port → vSwitch → VNF → vSwitch → physical port
237 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
238 .. code-block:: console
241 +---------------------------------------------------+ |
243 | +-------------------------------------------+ | |
244 | | Application | | |
245 | +-------------------------------------------+ | |
249 | +---------------+ +---------------+ | |
250 | | logical port 0| | logical port 1| | |
251 +---+---------------+-----------+---------------+---+ _|
255 +---+---------------+----------+---------------+---+ |
256 | | logical port 0| | logical port 1| | |
257 | +---------------+ +---------------+ | |
261 | +--------------+ +--------------+ | |
262 | | phy port | vSwitch | phy port | | |
263 +---+--------------+------------+--------------+---+ _|
267 +--------------------------------------------------+
269 | traffic generator |
271 +--------------------------------------------------+
275 Physical port → vSwitch → VNF → vSwitch → VNF → vSwitch → physical port
276 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
278 .. code-block:: console
281 +----------------------+ +----------------------+ |
282 | Guest 1 | | Guest 2 | |
283 | +---------------+ | | +---------------+ | |
284 | | Application | | | | Application | | |
285 | +---------------+ | | +---------------+ | |
287 | | v | | | v | | Guests
288 | +---------------+ | | +---------------+ | |
289 | | logical ports | | | | logical ports | | |
290 | | 0 1 | | | | 0 1 | | |
291 +---+---------------+--+ +---+---------------+--+ _|
295 +---+---------------+---------+---------------+--+ |
296 | | 0 1 | | 3 4 | | |
297 | | logical ports | | logical ports | | |
298 | +---------------+ +---------------+ | |
300 | | L-----------------+ v | |
301 | +--------------+ +--------------+ | |
302 | | phy ports | vSwitch | phy ports | | |
303 +---+--------------+----------+--------------+---+ _|
307 +--------------------------------------------------+
309 | traffic generator |
311 +--------------------------------------------------+
315 Physical port → VNF → vSwitch → VNF → physical port
316 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
318 .. code-block:: console
321 +----------------------+ +----------------------+ |
322 | Guest 1 | | Guest 2 | |
323 |+-------------------+ | | +-------------------+| |
324 || Application | | | | Application || |
325 |+-------------------+ | | +-------------------+| |
326 | ^ | | | ^ | | | Guests
328 |+-------------------+ | | +-------------------+| |
329 || logical ports | | | | logical ports || |
330 || 0 1 | | | | 0 1 || |
331 ++--------------------++ ++--------------------++ _|
333 (PCI passthrough) | | (PCI passthrough)
335 +--------++------------+-+------------++---------+ |
336 | | || 0 | | 1 || | | |
337 | | ||logical port| |logical port|| | | |
338 | | |+------------+ +------------+| | | |
340 | | | L-----------------+ | | | |
342 | | | vSwitch | | | |
343 | | +-----------------------------+ | | |
346 | +--------------+ +--------------+ | |
347 | | phy port/VF | | phy port/VF | | |
348 +-+--------------+--------------+--------------+-+ _|
352 +--------------------------------------------------+
354 | traffic generator |
356 +--------------------------------------------------+
360 Physical port → vSwitch → VNF
361 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
363 .. code-block:: console
366 +---------------------------------------------------+ |
368 | +-------------------------------------------+ | |
369 | | Application | | |
370 | +-------------------------------------------+ | |
374 | +---------------+ | |
375 | | logical port 0| | |
376 +---+---------------+-------------------------------+ _|
380 +---+---------------+------------------------------+ |
381 | | logical port 0| | |
382 | +---------------+ | |
386 | +--------------+ | |
387 | | phy port | vSwitch | |
388 +---+--------------+------------ -------------- ---+ _|
392 +--------------------------------------------------+
394 | traffic generator |
396 +--------------------------------------------------+
400 VNF → vSwitch → physical port
401 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
403 .. code-block:: console
406 +---------------------------------------------------+ |
408 | +-------------------------------------------+ | |
409 | | Application | | |
410 | +-------------------------------------------+ | |
414 | +---------------+ | |
415 | | logical port | | |
416 +-------------------------------+---------------+---+ _|
420 +------------------------------+---------------+---+ |
421 | | logical port | | |
422 | +---------------+ | |
426 | +--------------+ | |
427 | vSwitch | phy port | | |
428 +-------------------------------+--------------+---+ _|
432 +--------------------------------------------------+
434 | traffic generator |
436 +--------------------------------------------------+
440 VNF → vSwitch → VNF → vSwitch
441 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
443 .. code-block:: console
446 +-------------------------+ +-------------------------+ |
447 | Guest 1 | | Guest 2 | |
448 | +-----------------+ | | +-----------------+ | |
449 | | Application | | | | Application | | |
450 | +-----------------+ | | +-----------------+ | |
454 | +---------------+ | | +---------------+ | |
455 | | logical port 0| | | | logical port 0| | |
456 +-----+---------------+---+ +---+---------------+-----+ _|
460 +----+---------------+------------+---------------+-----+ |
461 | | port 0 | | port 1 | | |
462 | +---------------+ +---------------+ | |
465 | +--------------------+ | |
468 +-------------------------------------------------------+ _|
472 HOST 1(Physical port → virtual switch → VNF → virtual switch → Physical port)
473 → HOST 2(Physical port → virtual switch → VNF → virtual switch → Physical port)
475 HOST 1 (PVP) → HOST 2 (PVP)
476 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
478 .. code-block:: console
481 +----------------------+ +----------------------+ |
482 | Guest 1 | | Guest 2 | |
483 | +---------------+ | | +---------------+ | |
484 | | Application | | | | Application | | |
485 | +---------------+ | | +---------------+ | |
487 | | v | | | v | | Guests
488 | +---------------+ | | +---------------+ | |
489 | | logical ports | | | | logical ports | | |
490 | | 0 1 | | | | 0 1 | | |
491 +---+---------------+--+ +---+---------------+--+ _|
495 +---+---------------+--+ +---+---------------+--+ |
496 | | 0 1 | | | | 3 4 | | |
497 | | logical ports | | | | logical ports | | |
498 | +---------------+ | | +---------------+ | |
499 | ^ | | | ^ | | | Hosts
501 | +--------------+ | | +--------------+ | |
502 | | phy ports | | | | phy ports | | |
503 +---+--------------+---+ +---+--------------+---+ _|
505 | +-----------------+ |
507 +--------------------------------------------------+
509 | traffic generator |
511 +--------------------------------------------------+
515 **Note:** For tests where the traffic generator and/or measurement
516 receiver are implemented on VM and connected to the virtual switch
517 through vNIC, the issues of shared resources and interactions between
518 the measurement devices and the device under test must be considered.
520 **Note:** Some RFC 2889 tests require a full-mesh sending and receiving
521 pattern involving more than two ports. This possibility is illustrated in the
522 Physical port → vSwitch → VNF → vSwitch → VNF → vSwitch → physical port
523 diagram above (with 2 sending and 2 receiving ports, though all ports
524 could be used bi-directionally).
526 **Note:** When Deployment Scenarios are used in RFC 2889 address learning
527 or cache capacity testing, an additional port from the vSwitch must be
528 connected to the test device. This port is used to listen for flooded
534 --------------------------
535 To establish the baseline performance of the virtual switch, tests would
536 initially be run with a simple workload in the VNF (the recommended
537 simple workload VNF would be `DPDK <http://www.dpdk.org/>`__'s testpmd
538 application forwarding packets in a VM or vloop\_vnf a simple kernel
539 module that forwards traffic between two network interfaces inside the
540 virtualized environment while bypassing the networking stack).
541 Subsequently, the tests would also be executed with a real Telco
542 workload running in the VNF, which would exercise the virtual switch in
543 the context of higher level Telco NFV use cases, and prove that its
544 underlying characteristics and behaviour can be measured and validated.
545 Suitable real Telco workload VNFs are yet to be identified.
549 Default Test Parameters
550 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
552 The following list identifies the default parameters for suite of
555 - Reference application: Simple forwarding or Open Source VNF.
556 - Frame size (bytes): 64, 128, 256, 512, 1024, 1280, 1518, 2K, 4k OR
557 Packet size based on use-case (e.g. RTP 64B, 256B) OR Mix of packet sizes as
558 maintained by the Functest project <https://wiki.opnfv.org/traffic_profile_management>.
559 - Reordering check: Tests should confirm that packets within a flow are
561 - Duplex: Unidirectional / Bidirectional. Default: Full duplex with
562 traffic transmitting in both directions, as network traffic generally
563 does not flow in a single direction. By default the data rate of
564 transmitted traffic should be the same in both directions, please
565 note that asymmetric traffic (e.g. downlink-heavy) tests will be
566 mentioned explicitly for the relevant test cases.
567 - Number of Flows: Default for non scalability tests is a single flow.
568 For scalability tests the goal is to test with maximum supported
569 flows but where possible will test up to 10 Million flows. Start with
570 a single flow and scale up. By default flows should be added
571 sequentially, tests that add flows simultaneously will explicitly
572 call out their flow addition behaviour. Packets are generated across
573 the flows uniformly with no burstiness. For multi-core tests should
574 consider the number of packet flows based on vSwitch/VNF multi-thread
575 implementation and behavior.
577 - Traffic Types: UDP, SCTP, RTP, GTP and UDP traffic.
578 - Deployment scenarios are:
579 - Physical → virtual switch → physical.
580 - Physical → virtual switch → VNF → virtual switch → physical.
581 - Physical → virtual switch → VNF → virtual switch → VNF → virtual
583 - Physical → VNF → virtual switch → VNF → physical.
584 - Physical → virtual switch → VNF.
585 - VNF → virtual switch → Physical.
586 - VNF → virtual switch → VNF.
588 Tests MUST have these parameters unless otherwise stated. **Test cases
589 with non default parameters will be stated explicitly**.
591 **Note**: For throughput tests unless stated otherwise, test
592 configurations should ensure that traffic traverses the installed flows
593 through the virtual switch, i.e. flows are installed and have an appropriate
594 time out that doesn't expire before packet transmission starts.
599 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
601 Virtual switches classify packets into flows by processing and matching
602 particular header fields in the packet/frame and/or the input port where
603 the packets/frames arrived. The vSwitch then carries out an action on
604 the group of packets that match the classification parameters. Thus a
605 flow is considered to be a sequence of packets that have a shared set of
606 header field values or have arrived on the same port and have the same
607 action applied to them. Performance results can vary based on the
608 parameters the vSwitch uses to match for a flow. The recommended flow
609 classification parameters for L3 vSwitch performance tests are: the
610 input port, the source IP address, the destination IP address and the
611 Ethernet protocol type field. It is essential to increase the flow
612 time-out time on a vSwitch before conducting any performance tests that
613 do not measure the flow set-up time. Normally the first packet of a
614 particular flow will install the flow in the vSwitch which adds an
615 additional latency, subsequent packets of the same flow are not subject
616 to this latency if the flow is already installed on the vSwitch.
621 ~~~~~~~~~~~~~~~~~~~~~
623 Tests will be assigned a priority in order to determine which tests
624 should be implemented immediately and which tests implementations
627 Priority can be of following types: - Urgent: Must be implemented
628 immediately. - High: Must be implemented in the next release. - Medium:
629 May be implemented after the release. - Low: May or may not be
637 The SUT should be configured to its "default" state. The
638 SUT's configuration or set-up must not change between tests in any way
639 other than what is required to do the test. All supported protocols must
640 be configured and enabled for each test set up.
645 ~~~~~~~~~~~~~~~~~~~~~~~~~~
647 The DUT should be configured with n ports where
648 n is a multiple of 2. Half of the ports on the DUT should be used as
649 ingress ports and the other half of the ports on the DUT should be used
650 as egress ports. Where a DUT has more than 2 ports, the ingress data
651 streams should be set-up so that they transmit packets to the egress
652 ports in sequence so that there is an even distribution of traffic
653 across ports. For example, if a DUT has 4 ports 0(ingress), 1(ingress),
654 2(egress) and 3(egress), the traffic stream directed at port 0 should
655 output a packet to port 2 followed by a packet to port 3. The traffic
656 stream directed at port 1 should also output a packet to port 2 followed
657 by a packet to port 3.
662 ~~~~~~~~~~~~~~~~~~~~~
664 **Frame formats Layer 2 (data link layer) protocols**
668 .. code-block:: console
670 +---------------------------+-----------+
671 | Ethernet Header | Payload | Check Sum |
672 +-----------------+---------+-----------+
673 |_________________|_________|___________|
674 14 Bytes 46 - 1500 4 Bytes
678 **Layer 3 (network layer) protocols**
682 .. code-block:: console
684 +-----------------+-----------+---------+-----------+
685 | Ethernet Header | IP Header | Payload | Checksum |
686 +-----------------+-----------+---------+-----------+
687 |_________________|___________|_________|___________|
688 14 Bytes 20 bytes 26 - 1480 4 Bytes
693 .. code-block:: console
695 +-----------------+-----------+---------+-----------+
696 | Ethernet Header | IP Header | Payload | Checksum |
697 +-----------------+-----------+---------+-----------+
698 |_________________|___________|_________|___________|
699 14 Bytes 40 bytes 26 - 1460 4 Bytes
702 **Layer 4 (transport layer) protocols**
708 .. code-block:: console
710 +-----------------+-----------+-----------------+---------+-----------+
711 | Ethernet Header | IP Header | Layer 4 Header | Payload | Checksum |
712 +-----------------+-----------+-----------------+---------+-----------+
713 |_________________|___________|_________________|_________|___________|
714 14 Bytes 40 bytes 20 Bytes 6 - 1460 4 Bytes
718 **Layer 5 (application layer) protocols**
723 .. code-block:: console
725 +-----------------+-----------+-----------------+---------+-----------+
726 | Ethernet Header | IP Header | Layer 4 Header | Payload | Checksum |
727 +-----------------+-----------+-----------------+---------+-----------+
728 |_________________|___________|_________________|_________|___________|
729 14 Bytes 20 bytes 20 Bytes >= 6 Bytes 4 Bytes
734 ~~~~~~~~~~~~~~~~~~~~~~~~~
735 There is a difference between an Ethernet frame,
736 an IP packet, and a UDP datagram. In the seven-layer OSI model of
737 computer networking, packet refers to a data unit at layer 3 (network
738 layer). The correct term for a data unit at layer 2 (data link layer) is
739 a frame, and at layer 4 (transport layer) is a segment or datagram.
741 Important concepts related to 10GbE performance are frame rate and
742 throughput. The MAC bit rate of 10GbE, defined in the IEEE standard 802
743 .3ae, is 10 billion bits per second. Frame rate is based on the bit rate
744 and frame format definitions. Throughput, defined in IETF RFC 1242, is
745 the highest rate at which the system under test can forward the offered
748 The frame rate for 10GbE is determined by a formula that divides the 10
749 billion bits per second by the preamble + frame length + inter-frame
752 The maximum frame rate is calculated using the minimum values of the
753 following parameters, as described in the IEEE 802 .3ae standard:
755 - Preamble: 8 bytes \* 8 = 64 bits
756 - Frame Length: 64 bytes (minimum) \* 8 = 512 bits
757 - Inter-frame Gap: 12 bytes (minimum) \* 8 = 96 bits
759 Therefore, Maximum Frame Rate (64B Frames)
760 = MAC Transmit Bit Rate / (Preamble + Frame Length + Inter-frame Gap)
761 = 10,000,000,000 / (64 + 512 + 96)
762 = 10,000,000,000 / 672
763 = 14,880,952.38 frame per second (fps)
767 System isolation and validation
768 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
770 A key consideration when conducting any sort of benchmark is trying to
771 ensure the consistency and repeatability of test results between runs.
772 When benchmarking the performance of a virtual switch there are many
773 factors that can affect the consistency of results. This section
774 describes these factors and the measures that can be taken to limit
775 their effects. In addition, this section will outline some system tests
776 to validate the platform and the VNF before conducting any vSwitch
779 **System Isolation:**
781 When conducting a benchmarking test on any SUT, it is essential to limit
782 (and if reasonable, eliminate) any noise that may interfere with the
783 accuracy of the metrics collected by the test. This noise may be
784 introduced by other hardware or software (OS, other applications), and
785 can result in significantly varying performance metrics being collected
786 between consecutive runs of the same test. In the case of characterizing
787 the performance of a virtual switch, there are a number of configuration
788 parameters that can help increase the repeatability and stability of
789 test results, including:
791 - OS/GRUB configuration:
793 - maxcpus = n where n >= 0; limits the kernel to using 'n'
794 processors. Only use exactly what you need.
795 - isolcpus: Isolate CPUs from the general scheduler. Isolate all
796 CPUs bar one which will be used by the OS.
797 - use taskset to affinitize the forwarding application and the VNFs
798 onto isolated cores. VNFs and the vSwitch should be allocated
799 their own cores, i.e. must not share the same cores. vCPUs for the
800 VNF should be affinitized to individual cores also.
801 - Limit the amount of background applications that are running and
802 set OS to boot to runlevel 3. Make sure to kill any unnecessary
803 system processes/daemons.
804 - Only enable hardware that you need to use for your test – to
805 ensure there are no other interrupts on the system.
806 - Configure NIC interrupts to only use the cores that are not
807 allocated to any other process (VNF/vSwitch).
809 - NUMA configuration: Any unused sockets in a multi-socket system
811 - CPU pinning: The vSwitch and the VNF should each be affinitized to
812 separate logical cores using a combination of maxcpus, isolcpus and
814 - BIOS configuration: BIOS should be configured for performance where
815 an explicit option exists, sleep states should be disabled, any
816 virtualization optimization technologies should be enabled, and
817 hyperthreading should also be enabled, turbo boost and overclocking
820 **System Validation:**
822 System validation is broken down into two sub-categories: Platform
823 validation and VNF validation. The validation test itself involves
824 verifying the forwarding capability and stability for the sub-system
825 under test. The rationale behind system validation is two fold. Firstly
826 to give a tester confidence in the stability of the platform or VNF that
827 is being tested; and secondly to provide base performance comparison
828 points to understand the overhead introduced by the virtual switch.
830 * Benchmark platform forwarding capability: This is an OPTIONAL test
831 used to verify the platform and measure the base performance (maximum
832 forwarding rate in fps and latency) that can be achieved by the
833 platform without a vSwitch or a VNF. The following diagram outlines
834 the set-up for benchmarking Platform forwarding capability:
836 .. code-block:: console
839 +--------------------------------------------------+ |
840 | +------------------------------------------+ | |
842 | | l2fw or DPDK L2FWD app | | Host
844 | +------------------------------------------+ | |
846 +---+------------------------------------------+---+ __|
850 +--------------------------------------------------+
852 | traffic generator |
854 +--------------------------------------------------+
856 * Benchmark VNF forwarding capability: This test is used to verify
857 the VNF and measure the base performance (maximum forwarding rate in
858 fps and latency) that can be achieved by the VNF without a vSwitch.
859 The performance metrics collected by this test will serve as a key
860 comparison point for NIC passthrough technologies and vSwitches. VNF
861 in this context refers to the hypervisor and the VM. The following
862 diagram outlines the set-up for benchmarking VNF forwarding
865 .. code-block:: console
868 +--------------------------------------------------+ |
869 | +------------------------------------------+ | |
873 | +------------------------------------------+ | |
874 | | Passthrough/SR-IOV | | Host
875 | +------------------------------------------+ | |
877 +---+------------------------------------------+---+ __|
881 +--------------------------------------------------+
883 | traffic generator |
885 +--------------------------------------------------+
888 **Methodology to benchmark Platform/VNF forwarding capability**
891 The recommended methodology for the platform/VNF validation and
892 benchmark is: - Run `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
893 Maximum Forwarding Rate test, this test will produce maximum
894 forwarding rate and latency results that will serve as the
895 expected values. These expected values can be used in
896 subsequent steps or compared with in subsequent validation tests. -
897 Transmit bidirectional traffic at line rate/max forwarding rate
898 (whichever is higher) for at least 72 hours, measure throughput (fps)
899 and latency. - Note: Traffic should be bidirectional. - Establish a
900 baseline forwarding rate for what the platform can achieve. - Additional
901 validation: After the test has completed for 72 hours run bidirectional
902 traffic at the maximum forwarding rate once more to see if the system is
903 still functional and measure throughput (fps) and latency. Compare the
904 measure the new obtained values with the expected values.
906 **NOTE 1**: How the Platform is configured for its forwarding capability
907 test (BIOS settings, GRUB configuration, runlevel...) is how the
908 platform should be configured for every test after this
910 **NOTE 2**: How the VNF is configured for its forwarding capability test
911 (# of vCPUs, vNICs, Memory, affinitization…) is how it should be
912 configured for every test that uses a VNF after this.
916 RFCs for testing virtual switch performance
917 --------------------------------------------------
919 The starting point for defining the suite of tests for benchmarking the
920 performance of a virtual switch is to take existing RFCs and standards
921 that were designed to test their physical counterparts and adapting them
922 for testing virtual switches. The rationale behind this is to establish
923 a fair comparison between the performance of virtual and physical
924 switches. This section outlines the RFCs that are used by this
929 RFC 1242 Benchmarking Terminology for Network Interconnection
930 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
931 Devices RFC 1242 defines the terminology that is used in describing
932 performance benchmarking tests and their results. Definitions and
933 discussions covered include: Back-to-back, bridge, bridge/router,
934 constant load, data link frame size, frame loss rate, inter frame gap,
935 latency, and many more.
939 RFC 2544 Benchmarking Methodology for Network Interconnect Devices
940 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
941 RFC 2544 outlines a benchmarking methodology for network Interconnect
942 Devices. The methodology results in performance metrics such as latency,
943 frame loss percentage, and maximum data throughput.
945 In this document network “throughput” (measured in millions of frames
946 per second) is based on RFC 2544, unless otherwise noted. Frame size
947 refers to Ethernet frames ranging from smallest frames of 64 bytes to
948 largest frames of 9K bytes.
952 1. Throughput test defines the maximum number of frames per second
953 that can be transmitted without any error.
955 2. Latency test measures the time required for a frame to travel from
956 the originating device through the network to the destination device.
957 Please note that RFC2544 Latency measurement will be superseded with
958 a measurement of average latency over all successfully transferred
961 3. Frame loss test measures the network’s
962 response in overload conditions - a critical indicator of the
963 network’s ability to support real-time applications in which a
964 large amount of frame loss will rapidly degrade service quality.
966 4. Burst test assesses the buffering capability of a virtual switch. It
967 measures the maximum number of frames received at full line rate
968 before a frame is lost. In carrier Ethernet networks, this
969 measurement validates the excess information rate (EIR) as defined in
972 5. System recovery to characterize speed of recovery from an overload
975 6. Reset to characterize speed of recovery from device or software
976 reset. This type of test has been updated by `RFC6201
977 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__ as such,
978 the methodology defined by this specification will be that of RFC 6201.
980 Although not included in the defined RFC 2544 standard, another crucial
981 measurement in Ethernet networking is packet delay variation. The
982 definition set out by this specification comes from
983 `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__.
987 RFC 2285 Benchmarking Terminology for LAN Switching Devices
988 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
989 RFC 2285 defines the terminology that is used to describe the
990 terminology for benchmarking a LAN switching device. It extends RFC
991 1242 and defines: DUTs, SUTs, Traffic orientation and distribution,
992 bursts, loads, forwarding rates, etc.
996 RFC 2889 Benchmarking Methodology for LAN Switching
997 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
998 RFC 2889 outlines a benchmarking methodology for LAN switching, it
999 extends RFC 2544. The outlined methodology gathers performance
1000 metrics for forwarding, congestion control, latency, address handling
1001 and finally filtering.
1005 RFC 3918 Methodology for IP Multicast Benchmarking
1006 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1007 RFC 3918 outlines a methodology for IP Multicast benchmarking.
1011 RFC 4737 Packet Reordering Metrics
1012 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1013 RFC 4737 describes metrics for identifying and counting re-ordered
1014 packets within a stream, and metrics to measure the extent each
1015 packet has been re-ordered.
1019 RFC 5481 Packet Delay Variation Applicability Statement
1020 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1021 RFC 5481 defined two common, but different forms of delay variation
1022 metrics, and compares the metrics over a range of networking
1023 circumstances and tasks. The most suitable form for vSwitch
1024 benchmarking is the "PDV" form.
1028 RFC 6201 Device Reset Characterization
1029 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1030 RFC 6201 extends the methodology for characterizing the speed of
1031 recovery of the DUT from device or software reset described in RFC
1036 Details of the Test Report
1037 ---------------------------------
1039 There are a number of parameters related to the system, DUT and tests
1040 that can affect the repeatability of a test results and should be
1041 recorded. In order to minimise the variation in the results of a test,
1042 it is recommended that the test report includes the following information:
1044 - Hardware details including:
1047 - Processor details.
1048 - Memory information (see below)
1049 - Number of enabled cores.
1050 - Number of cores used for the test.
1051 - Number of physical NICs, as well as their details (manufacturer,
1052 versions, type and the PCI slot they are plugged into).
1053 - NIC interrupt configuration.
1054 - BIOS version, release date and any configurations that were
1057 - Software details including:
1059 - OS version (for host and VNF)
1060 - Kernel version (for host and VNF)
1061 - GRUB boot parameters (for host and VNF).
1062 - Hypervisor details (Type and version).
1063 - Selected vSwitch, version number or commit id used.
1064 - vSwitch launch command line if it has been parameterised.
1065 - Memory allocation to the vSwitch – which NUMA node it is using,
1066 and how many memory channels.
1067 - Where the vswitch is built from source: compiler details including
1068 versions and the flags that were used to compile the vSwitch.
1069 - DPDK or any other SW dependency version number or commit id used.
1070 - Memory allocation to a VM - if it's from Hugpages/elsewhere.
1071 - VM storage type: snapshot/independent persistent/independent
1074 - Number of Virtual NICs (vNICs), versions, type and driver.
1075 - Number of virtual CPUs and their core affinity on the host.
1076 - Number vNIC interrupt configuration.
1077 - Thread affinitization for the applications (including the vSwitch
1078 itself) on the host.
1079 - Details of Resource isolation, such as CPUs designated for
1080 Host/Kernel (isolcpu) and CPUs designated for specific processes
1099 - Traffic Information:
1101 - Traffic type - UDP, TCP, IMIX / Other.
1104 - Deployment Scenario.
1106 **Note**: Tests that require additional parameters to be recorded will
1107 explicitly specify this.
1109 .. _TestIdentification:
1114 =========================
1119 ----------------------
1120 The following tests aim to determine the maximum forwarding rate that
1121 can be achieved with a virtual switch. The list is not exhaustive but
1122 should indicate the type of tests that should be required. It is
1123 expected that more will be added.
1127 Test ID: LTD.Throughput.RFC2544.PacketLossRatio
1128 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1129 **Title**: RFC 2544 X% packet loss ratio Throughput and Latency Test
1131 **Prerequisite Test**: N/A
1137 This test determines the DUT's maximum forwarding rate with X% traffic
1138 loss for a constant load (fixed length frames at a fixed interval time).
1139 The default loss percentages to be tested are: - X = 0% - X = 10^-7%
1141 Note: Other values can be tested if required by the user.
1143 The selected frame sizes are those previously defined under `Default
1144 Test Parameters <#DefaultParams>`__. The test can also be used to
1145 determine the average latency of the traffic.
1147 Under the `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__
1148 test methodology, the test duration will
1149 include a number of trials; each trial should run for a minimum period
1150 of 60 seconds. A binary search methodology must be applied for each
1151 trial to obtain the final result.
1153 **Expected Result**: At the end of each trial, the presence or absence
1154 of loss determines the modification of offered load for the next trial,
1155 converging on a maximum rate, or
1156 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ Throughput with X%
1158 The Throughput load is re-used in related
1159 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ tests and other
1162 **Metrics Collected**:
1164 The following are the metrics collected for this test:
1166 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
1167 the DUT for each frame size with X% packet loss.
1168 - The average latency of the traffic flow when passing through the DUT
1169 (if testing for latency, note that this average is different from the
1170 test specified in Section 26.3 of
1171 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
1172 - CPU and memory utilization may also be collected as part of this
1173 test, to determine the vSwitch's performance footprint on the system.
1177 Test ID: LTD.Throughput.RFC2544.PacketLossRatioFrameModification
1178 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1179 **Title**: RFC 2544 X% packet loss Throughput and Latency Test with
1182 **Prerequisite Test**: N/A
1188 This test determines the DUT's maximum forwarding rate with X% traffic
1189 loss for a constant load (fixed length frames at a fixed interval time).
1190 The default loss percentages to be tested are: - X = 0% - X = 10^-7%
1192 Note: Other values can be tested if required by the user.
1194 The selected frame sizes are those previously defined under `Default
1195 Test Parameters <#DefaultParams>`__. The test can also be used to
1196 determine the average latency of the traffic.
1198 Under the `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__
1199 test methodology, the test duration will
1200 include a number of trials; each trial should run for a minimum period
1201 of 60 seconds. A binary search methodology must be applied for each
1202 trial to obtain the final result.
1204 During this test, the DUT must perform the following operations on the
1207 - Perform packet parsing on the DUT's ingress port.
1208 - Perform any relevant address look-ups on the DUT's ingress ports.
1209 - Modify the packet header before forwarding the packet to the DUT's
1210 egress port. Packet modifications include:
1212 - Modifying the Ethernet source or destination MAC address.
1213 - Modifying/adding a VLAN tag. (**Recommended**).
1214 - Modifying/adding a MPLS tag.
1215 - Modifying the source or destination ip address.
1216 - Modifying the TOS/DSCP field.
1217 - Modifying the source or destination ports for UDP/TCP/SCTP.
1218 - Modifying the TTL.
1220 **Expected Result**: The Packet parsing/modifications require some
1221 additional degree of processing resource, therefore the
1222 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__
1223 Throughput is expected to be somewhat lower than the Throughput level
1224 measured without additional steps. The reduction is expected to be
1225 greatest on tests with the smallest packet sizes (greatest header
1228 **Metrics Collected**:
1230 The following are the metrics collected for this test:
1232 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
1233 the DUT for each frame size with X% packet loss and packet
1234 modification operations being performed by the DUT.
1235 - The average latency of the traffic flow when passing through the DUT
1236 (if testing for latency, note that this average is different from the
1237 test specified in Section 26.3 of
1238 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
1239 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1240 PDV form of delay variation on the traffic flow,
1241 using the 99th percentile.
1242 - CPU and memory utilization may also be collected as part of this
1243 test, to determine the vSwitch's performance footprint on the system.
1247 Test ID: LTD.Throughput.RFC2544.Profile
1248 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1249 **Title**: RFC 2544 Throughput and Latency Profile
1251 **Prerequisite Test**: N/A
1257 This test reveals how throughput and latency degrades as the offered
1258 rate varies in the region of the DUT's maximum forwarding rate as
1259 determined by LTD.Throughput.RFC2544.PacketLossRatio (0% Packet Loss).
1260 For example it can be used to determine if the degradation of throughput
1261 and latency as the offered rate increases is slow and graceful or sudden
1264 The selected frame sizes are those previously defined under `Default
1265 Test Parameters <#DefaultParams>`__.
1267 The offered traffic rate is described as a percentage delta with respect
1268 to the DUT's RFC 2544 Throughput as determined by
1269 LTD.Throughput.RFC2544.PacketLoss Ratio (0% Packet Loss case). A delta
1270 of 0% is equivalent to an offered traffic rate equal to the RFC 2544
1271 Throughput; A delta of +50% indicates an offered rate half-way
1272 between the Throughput and line-rate, whereas a delta of
1273 -50% indicates an offered rate of half the maximum rate. Therefore the
1274 range of the delta figure is natuarlly bounded at -100% (zero offered
1275 traffic) and +100% (traffic offered at line rate).
1277 The following deltas to the maximum forwarding rate should be applied:
1279 - -50%, -10%, 0%, +10% & +50%
1281 **Expected Result**: For each packet size a profile should be produced
1282 of how throughput and latency vary with offered rate.
1284 **Metrics Collected**:
1286 The following are the metrics collected for this test:
1288 - The forwarding rate in Frames Per Second (FPS) and Mbps of the DUT
1289 for each delta to the maximum forwarding rate and for each frame
1291 - The average latency for each delta to the maximum forwarding rate and
1292 for each frame size.
1293 - CPU and memory utilization may also be collected as part of this
1294 test, to determine the vSwitch's performance footprint on the system.
1295 - Any failures experienced (for example if the vSwitch crashes, stops
1296 processing packets, restarts or becomes unresponsive to commands)
1297 when the offered load is above Maximum Throughput MUST be recorded
1298 and reported with the results.
1302 Test ID: LTD.Throughput.RFC2544.SystemRecoveryTime
1303 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1304 **Title**: RFC 2544 System Recovery Time Test
1306 **Prerequisite Test** LTD.Throughput.RFC2544.PacketLossRatio
1312 The aim of this test is to determine the length of time it takes the DUT
1313 to recover from an overload condition for a constant load (fixed length
1314 frames at a fixed interval time). The selected frame sizes are those
1315 previously defined under `Default Test Parameters <#DefaultParams>`__,
1316 traffic should be sent to the DUT under normal conditions. During the
1317 duration of the test and while the traffic flows are passing though the
1318 DUT, at least one situation leading to an overload condition for the DUT
1319 should occur. The time from the end of the overload condition to when
1320 the DUT returns to normal operations should be measured to determine
1321 recovery time. Prior to overloading the DUT, one should record the
1322 average latency for 10,000 packets forwarded through the DUT.
1324 The overload condition SHOULD be to transmit traffic at a very high
1325 frame rate to the DUT (150% of the maximum 0% packet loss rate as
1326 determined by LTD.Throughput.RFC2544.PacketLossRatio or line-rate
1327 whichever is lower), for at least 60 seconds, then reduce the frame rate
1328 to 75% of the maximum 0% packet loss rate. A number of time-stamps
1329 should be recorded: - Record the time-stamp at which the frame rate was
1330 reduced and record a second time-stamp at the time of the last frame
1331 lost. The recovery time is the difference between the two timestamps. -
1332 Record the average latency for 10,000 frames after the last frame loss
1333 and continue to record average latency measurements for every 10,000
1334 frames, when latency returns to within 10% of pre-overload levels record
1337 **Expected Result**:
1339 **Metrics collected**
1341 The following are the metrics collected for this test:
1343 - The length of time it takes the DUT to recover from an overload
1345 - The length of time it takes the DUT to recover the average latency to
1346 pre-overload conditions.
1348 **Deployment scenario**:
1350 - Physical → virtual switch → physical.
1354 Test ID: LTD.Throughput.RFC2544.BackToBackFrames
1355 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1356 **Title**: RFC2544 Back To Back Frames Test
1358 **Prerequisite Test**: N
1364 The aim of this test is to characterize the ability of the DUT to
1365 process back-to-back frames. For each frame size previously defined
1366 under `Default Test Parameters <#DefaultParams>`__, a burst of traffic
1367 is sent to the DUT with the minimum inter-frame gap between each frame.
1368 If the number of received frames equals the number of frames that were
1369 transmitted, the burst size should be increased and traffic is sent to
1370 the DUT again. The value measured is the back-to-back value, that is the
1371 maximum burst size the DUT can handle without any frame loss. Please note
1372 a trial must run for a minimum of 2 seconds and should be repeated 50
1373 times (at a minimum).
1375 **Expected Result**:
1377 Tests of back-to-back frames with physical devices have produced
1378 unstable results in some cases. All tests should be repeated in multiple
1379 test sessions and results stability should be examined.
1381 **Metrics collected**
1383 The following are the metrics collected for this test:
1385 - The average back-to-back value across the trials, which is
1386 the number of frames in the longest burst that the DUT will
1387 handle without the loss of any frames.
1388 - CPU and memory utilization may also be collected as part of this
1389 test, to determine the vSwitch's performance footprint on the system.
1391 **Deployment scenario**:
1393 - Physical → virtual switch → physical.
1397 Test ID: LTD.Throughput.RFC2889.MaxForwardingRateSoak
1398 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1399 **Title**: RFC 2889 X% packet loss Max Forwarding Rate Soak Test
1401 **Prerequisite Test** LTD.Throughput.RFC2544.PacketLossRatio
1407 The aim of this test is to understand the Max Forwarding Rate stability
1408 over an extended test duration in order to uncover any outliers. To allow
1409 for an extended test duration, the test should ideally run for 24 hours
1410 or, if this is not possible, for at least 6 hours. For this test, each frame
1411 size must be sent at the highest Throughput rate with X% packet loss, as
1412 determined in the prerequisite test. The default loss percentages to be
1413 tested are: - X = 0% - X = 10^-7%
1415 Note: Other values can be tested if required by the user.
1417 **Expected Result**:
1419 **Metrics Collected**:
1421 The following are the metrics collected for this test:
1423 - Max Forwarding Rate stability of the DUT.
1425 - This means reporting the number of packets lost per time interval
1426 and reporting any time intervals with packet loss. The
1427 `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
1428 Forwarding Rate shall be measured in each interval.
1429 An interval of 60s is suggested.
1431 - CPU and memory utilization may also be collected as part of this
1432 test, to determine the vSwitch's performance footprint on the system.
1433 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1434 PDV form of delay variation on the traffic flow,
1435 using the 99th percentile.
1439 Test ID: LTD.Throughput.RFC2889.MaxForwardingRateSoakFrameModification
1440 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1441 **Title**: RFC 2889 Max Forwarding Rate Soak Test with Frame Modification
1443 **Prerequisite Test**:
1444 LTD.Throughput.RFC2544.PacketLossRatioFrameModification (0% Packet Loss)
1450 The aim of this test is to understand the Max Forwarding Rate stability over an
1451 extended test duration in order to uncover any outliers. To allow for an
1452 extended test duration, the test should ideally run for 24 hours or, if
1453 this is not possible, for at least 6 hour. For this test, each frame
1454 size must be sent at the highest Throughput rate with 0% packet loss, as
1455 determined in the prerequisite test.
1457 During this test, the DUT must perform the following operations on the
1460 - Perform packet parsing on the DUT's ingress port.
1461 - Perform any relevant address look-ups on the DUT's ingress ports.
1462 - Modify the packet header before forwarding the packet to the DUT's
1463 egress port. Packet modifications include:
1465 - Modifying the Ethernet source or destination MAC address.
1466 - Modifying/adding a VLAN tag (**Recommended**).
1467 - Modifying/adding a MPLS tag.
1468 - Modifying the source or destination ip address.
1469 - Modifying the TOS/DSCP field.
1470 - Modifying the source or destination ports for UDP/TCP/SCTP.
1471 - Modifying the TTL.
1473 **Expected Result**:
1475 **Metrics Collected**:
1477 The following are the metrics collected for this test:
1479 - Max Forwarding Rate stability of the DUT.
1481 - This means reporting the number of packets lost per time interval
1482 and reporting any time intervals with packet loss. The
1483 `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
1484 Forwarding Rate shall be measured in each interval.
1485 An interval of 60s is suggested.
1487 - CPU and memory utilization may also be collected as part of this
1488 test, to determine the vSwitch's performance footprint on the system.
1489 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1490 PDV form of delay variation on the traffic flow, using the 99th
1495 Test ID: LTD.Throughput.RFC6201.ResetTime
1496 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1497 **Title**: RFC 6201 Reset Time Test
1499 **Prerequisite Test**: N/A
1505 The aim of this test is to determine the length of time it takes the DUT
1506 to recover from a reset.
1508 Two reset methods are defined - planned and unplanned. A planned reset
1509 requires stopping and restarting the virtual switch by the usual
1510 'graceful' method defined by it's documentation. An unplanned reset
1511 requires simulating a fatal internal fault in the virtual switch - for
1512 example by using kill -SIGKILL on a Linux environment.
1514 Both reset methods SHOULD be exercised.
1516 For each frame size previously defined under `Default Test
1517 Parameters <#DefaultParams>`__, traffic should be sent to the DUT under
1518 normal conditions. During the duration of the test and while the traffic
1519 flows are passing through the DUT, the DUT should be reset and the Reset
1520 time measured. The Reset time is the total time that a device is
1521 determined to be out of operation and includes the time to perform the
1522 reset and the time to recover from it (cf. `RFC6201
1523 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__).
1525 `RFC6201 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__ defines two methods
1526 to measure the Reset time:
1528 - Frame-Loss Method: which requires the monitoring of the number of
1529 lost frames and calculates the Reset time based on the number of
1530 frames lost and the offered rate according to the following
1533 .. code-block:: console
1535 Frames_lost (packets)
1536 Reset_time = -------------------------------------
1537 Offered_rate (packets per second)
1539 - Timestamp Method: which measures the time from which the last frame
1540 is forwarded from the DUT to the time the first frame is forwarded
1541 after the reset. This involves time-stamping all transmitted frames
1542 and recording the timestamp of the last frame that was received prior
1543 to the reset and also measuring the timestamp of the first frame that
1544 is received after the reset. The Reset time is the difference between
1545 these two timestamps.
1547 According to `RFC6201 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__ the
1548 choice of method depends on the test tool's capability; the Frame-Loss
1549 method SHOULD be used if the test tool supports:
1551 * Counting the number of lost frames per stream.
1552 * Transmitting test frame despite the physical link status.
1554 whereas the Timestamp method SHOULD be used if the test tool supports:
1555 * Timestamping each frame.
1556 * Monitoring received frame's timestamp.
1557 * Transmitting frames only if the physical link status is up.
1559 **Expected Result**:
1561 **Metrics collected**
1563 The following are the metrics collected for this test:
1565 * Average Reset Time over the number of trials performed.
1567 Results of this test should include the following information:
1569 * The reset method used.
1570 * Throughput in Fps and Mbps.
1571 * Average Frame Loss over the number of trials performed.
1572 * Average Reset Time in milliseconds over the number of trials performed.
1573 * Number of trials performed.
1574 * Protocol: IPv4, IPv6, MPLS, etc.
1575 * Frame Size in Octets
1576 * Port Media: Ethernet, Gigabit Ethernet (GbE), etc.
1577 * Port Speed: 10 Gbps, 40 Gbps etc.
1578 * Interface Encapsulation: Ethernet, Ethernet VLAN, etc.
1580 **Deployment scenario**:
1582 * Physical → virtual switch → physical.
1586 Test ID: LTD.Throughput.RFC2889.MaxForwardingRate
1587 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1588 **Title**: RFC2889 Forwarding Rate Test
1590 **Prerequisite Test**: LTD.Throughput.RFC2544.PacketLossRatio
1596 This test measures the DUT's Max Forwarding Rate when the Offered Load
1597 is varied between the throughput and the Maximum Offered Load for fixed
1598 length frames at a fixed time interval. The selected frame sizes are
1599 those previously defined under `Default Test
1600 Parameters <#DefaultParams>`__. The throughput is the maximum offered
1601 load with 0% frame loss (measured by the prerequisite test), and the
1602 Maximum Offered Load (as defined by
1603 `RFC2285 <https://www.rfc-editor.org/rfc/rfc2285.txt>`__) is *"the highest
1604 number of frames per second that an external source can transmit to a
1605 DUT/SUT for forwarding to a specified output interface or interfaces"*.
1607 Traffic should be sent to the DUT at a particular rate (TX rate)
1608 starting with TX rate equal to the throughput rate. The rate of
1609 successfully received frames at the destination counted (in FPS). If the
1610 RX rate is equal to the TX rate, the TX rate should be increased by a
1611 fixed step size and the RX rate measured again until the Max Forwarding
1614 The trial duration for each iteration should last for the period of time
1615 needed for the system to reach steady state for the frame size being
1616 tested. Under `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
1617 (Sec. 5.6.3.1) test methodology, the test
1618 duration should run for a minimum period of 30 seconds, regardless
1619 whether the system reaches steady state before the minimum duration
1622 **Expected Result**: According to
1623 `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__ The Max Forwarding
1624 Rate is the highest forwarding rate of a DUT taken from an iterative set of
1625 forwarding rate measurements. The iterative set of forwarding rate measurements
1626 are made by setting the intended load transmitted from an external source and
1627 measuring the offered load (i.e what the DUT is capable of forwarding). If the
1628 Throughput == the Maximum Offered Load, it follows that Max Forwarding Rate is
1629 equal to the Maximum Offered Load.
1631 **Metrics Collected**:
1633 The following are the metrics collected for this test:
1635 - The Max Forwarding Rate for the DUT for each packet size.
1636 - CPU and memory utilization may also be collected as part of this
1637 test, to determine the vSwitch's performance footprint on the system.
1639 **Deployment scenario**:
1641 - Physical → virtual switch → physical. Note: Full mesh tests with
1642 multiple ingress and egress ports are a key aspect of RFC 2889
1643 benchmarks, and scenarios with both 2 and 4 ports should be tested.
1644 In any case, the number of ports used must be reported.
1648 Test ID: LTD.Throughput.RFC2889.ForwardPressure
1649 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1650 **Title**: RFC2889 Forward Pressure Test
1652 **Prerequisite Test**: LTD.Throughput.RFC2889.MaxForwardingRate
1658 The aim of this test is to determine if the DUT transmits frames with an
1659 inter-frame gap that is less than 12 bytes. This test overloads the DUT
1660 and measures the output for forward pressure. Traffic should be
1661 transmitted to the DUT with an inter-frame gap of 11 bytes, this will
1662 overload the DUT by 1 byte per frame. The forwarding rate of the DUT
1665 **Expected Result**: The forwarding rate should not exceed the maximum
1666 forwarding rate of the DUT collected by
1667 LTD.Throughput.RFC2889.MaxForwardingRate.
1669 **Metrics collected**
1671 The following are the metrics collected for this test:
1673 - Forwarding rate of the DUT in FPS or Mbps.
1674 - CPU and memory utilization may also be collected as part of this
1675 test, to determine the vSwitch's performance footprint on the system.
1677 **Deployment scenario**:
1679 - Physical → virtual switch → physical.
1683 Test ID: LTD.Throughput.RFC2889.ErrorFramesFiltering
1684 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1685 **Title**: RFC2889 Error Frames Filtering Test
1687 **Prerequisite Test**: N/A
1693 The aim of this test is to determine whether the DUT will propagate any
1694 erroneous frames it receives or whether it is capable of filtering out
1695 the erroneous frames. Traffic should be sent with erroneous frames
1696 included within the flow at random intervals. Illegal frames that must
1697 be tested include: - Oversize Frames. - Undersize Frames. - CRC Errored
1698 Frames. - Dribble Bit Errored Frames - Alignment Errored Frames
1700 The traffic flow exiting the DUT should be recorded and checked to
1701 determine if the erroneous frames where passed through the DUT.
1703 **Expected Result**: Broken frames are not passed!
1705 **Metrics collected**
1707 No Metrics are collected in this test, instead it determines:
1709 - Whether the DUT will propagate erroneous frames.
1710 - Or whether the DUT will correctly filter out any erroneous frames
1711 from traffic flow with out removing correct frames.
1713 **Deployment scenario**:
1715 - Physical → virtual switch → physical.
1719 Test ID: LTD.Throughput.RFC2889.BroadcastFrameForwarding
1720 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1721 **Title**: RFC2889 Broadcast Frame Forwarding Test
1723 **Prerequisite Test**: N
1729 The aim of this test is to determine the maximum forwarding rate of the
1730 DUT when forwarding broadcast traffic. For each frame previously defined
1731 under `Default Test Parameters <#DefaultParams>`__, the traffic should
1732 be set up as broadcast traffic. The traffic throughput of the DUT should
1735 The test should be conducted with at least 4 physical ports on the DUT.
1736 The number of ports used MUST be recorded.
1738 As broadcast involves forwarding a single incoming packet to several
1739 destinations, the latency of a single packet is defined as the average
1740 of the latencies for each of the broadcast destinations.
1742 The incoming packet is transmitted on each of the other physical ports,
1743 it is not transmitted on the port on which it was received. The test MAY
1744 be conducted using different broadcasting ports to uncover any
1745 performance differences.
1747 **Expected Result**:
1749 **Metrics collected**:
1751 The following are the metrics collected for this test:
1753 - The forwarding rate of the DUT when forwarding broadcast traffic.
1754 - The minimum, average & maximum packets latencies observed.
1756 **Deployment scenario**:
1758 - Physical → virtual switch 3x physical. In the Broadcast rate testing,
1759 four test ports are required. One of the ports is connected to the test
1760 device, so it can send broadcast frames and listen for miss-routed frames.
1764 Test ID: LTD.Throughput.RFC2544.WorstN-BestN
1765 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1766 **Title**: Modified RFC 2544 X% packet loss ratio Throughput and Latency Test
1768 **Prerequisite Test**: N/A
1774 This test determines the DUT's maximum forwarding rate with X% traffic
1775 loss for a constant load (fixed length frames at a fixed interval time).
1776 The default loss percentages to be tested are: X = 0%, X = 10^-7%
1778 Modified RFC 2544 throughput benchmarking methodology aims to quantify
1779 the throughput measurement variations observed during standard RFC 2544
1780 benchmarking measurements of virtual switches and VNFs. The RFC2544
1781 binary search algorithm is modified to use more samples per test trial
1782 to drive the binary search and yield statistically more meaningful
1783 results. This keeps the heart of the RFC2544 methodology, still relying
1784 on the binary search of throughput at specified loss tolerance, while
1785 providing more useful information about the range of results seen in
1786 testing. Instead of using a single traffic trial per iteration step,
1787 each traffic trial is repeated N times and the success/failure of the
1788 iteration step is based on these N traffic trials. Two types of revised
1789 tests are defined - *Worst-of-N* and *Best-of-N*.
1793 *Worst-of-N* indicates the lowest expected maximum throughput for (
1794 packet size, loss tolerance) when repeating the test.
1796 1. Repeat the same test run N times at a set packet rate, record each
1798 2. Take the WORST result (highest packet loss) out of N result samples,
1799 called the Worst-of-N sample.
1800 3. If Worst-of-N sample has loss less than the set loss tolerance, then
1801 the step is successful - increase the test traffic rate.
1802 4. If Worst-of-N sample has loss greater than the set loss tolerance
1803 then the step failed - decrease the test traffic rate.
1808 *Best-of-N* indicates the highest expected maximum throughput for (
1809 packet size, loss tolerance) when repeating the test.
1811 1. Repeat the same traffic run N times at a set packet rate, record
1813 2. Take the BEST result (least packet loss) out of N result samples,
1814 called the Best-of-N sample.
1815 3. If Best-of-N sample has loss less than the set loss tolerance, then
1816 the step is successful - increase the test traffic rate.
1817 4. If Best-of-N sample has loss greater than the set loss tolerance,
1818 then the step failed - decrease the test traffic rate.
1821 Performing both Worst-of-N and Best-of-N benchmark tests yields lower
1822 and upper bounds of expected maximum throughput under the operating
1823 conditions, giving a very good indication to the user of the
1824 deterministic performance range for the tested setup.
1826 **Expected Result**: At the end of each trial series, the presence or
1827 absence of loss determines the modification of offered load for the
1828 next trial series, converging on a maximum rate, or
1829 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ Throughput
1831 The Throughput load is re-used in related
1832 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ tests and other
1835 **Metrics Collected**:
1837 The following are the metrics collected for this test:
1839 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
1840 the DUT for each frame size with X% packet loss.
1841 - The average latency of the traffic flow when passing through the DUT
1842 (if testing for latency, note that this average is different from the
1843 test specified in Section 26.3 of
1844 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
1845 - Following may also be collected as part of this test, to determine
1846 the vSwitch's performance footprint on the system:
1847 - CPU core utilization.
1848 - CPU cache utilization.
1850 - System bus (QPI, PCI, ...) utilization.
1851 - CPU cycles consumed per packet.
1855 Packet Latency tests
1856 ---------------------------
1857 These tests will measure the store and forward latency as well as the packet
1858 delay variation for various packet types through the virtual switch. The
1859 following list is not exhaustive but should indicate the type of tests
1860 that should be required. It is expected that more will be added.
1864 Test ID: LTD.PacketLatency.InitialPacketProcessingLatency
1865 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1866 **Title**: Initial Packet Processing Latency
1868 **Prerequisite Test**: N/A
1874 In some virtual switch architectures, the first packets of a flow will
1875 take the system longer to process than subsequent packets in the flow.
1876 This test determines the latency for these packets. The test will
1877 measure the latency of the packets as they are processed by the
1878 flow-setup-path of the DUT. There are two methods for this test, a
1879 recommended method and a nalternative method that can be used if it is
1880 possible to disable the fastpath of the virtual switch.
1882 Recommended method: This test will send 64,000 packets to the DUT, each
1883 belonging to a different flow. Average packet latency will be determined
1884 over the 64,000 packets.
1886 Alternative method: This test will send a single packet to the DUT after
1887 a fixed interval of time. The time interval will be equivalent to the
1888 amount of time it takes for a flow to time out in the virtual switch
1889 plus 10%. Average packet latency will be determined over 1,000,000
1892 This test is intended only for non-learning virtual switches; For learning
1893 virtual switches use RFC2889.
1895 For this test, only unidirectional traffic is required.
1897 **Expected Result**: The average latency for the initial packet of all
1898 flows should be greater than the latency of subsequent traffic.
1900 **Metrics Collected**:
1902 The following are the metrics collected for this test:
1904 - Average latency of the initial packets of all flows that are
1905 processed by the DUT.
1907 **Deployment scenario**:
1909 - Physical → Virtual Switch → Physical.
1913 Test ID: LTD.PacketDelayVariation.RFC3393.Soak
1914 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1915 **Title**: Packet Delay Variation Soak Test
1917 **Prerequisite Tests**: LTD.Throughput.RFC2544.PacketLossRatio (0% Packet Loss)
1923 The aim of this test is to understand the distribution of packet delay
1924 variation for different frame sizes over an extended test duration and
1925 to determine if there are any outliers. To allow for an extended test
1926 duration, the test should ideally run for 24 hours or, if this is not
1927 possible, for at least 6 hour. For this test, each frame size must be
1928 sent at the highest possible throughput with 0% packet loss, as
1929 determined in the prerequisite test.
1931 **Expected Result**:
1933 **Metrics Collected**:
1935 The following are the metrics collected for this test:
1937 - The packet delay variation value for traffic passing through the DUT.
1938 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1939 PDV form of delay variation on the traffic flow,
1940 using the 99th percentile, for each 60s interval during the test.
1941 - CPU and memory utilization may also be collected as part of this
1942 test, to determine the vSwitch's performance footprint on the system.
1947 ------------------------
1948 The general aim of these tests is to understand the impact of large flow
1949 table size and flow lookups on throughput. The following list is not
1950 exhaustive but should indicate the type of tests that should be required.
1951 It is expected that more will be added.
1955 Test ID: LTD.Scalability.RFC2544.0PacketLoss
1956 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1957 **Title**: RFC 2544 0% loss Scalability throughput test
1959 **Prerequisite Test**: LTD.Throughput.RFC2544.PacketLossRatio, IF the
1960 delta Throughput between the single-flow RFC2544 test and this test with
1961 a variable number of flows is desired.
1967 The aim of this test is to measure how throughput changes as the number
1968 of flows in the DUT increases. The test will measure the throughput
1969 through the fastpath, as such the flows need to be installed on the DUT
1970 before passing traffic.
1972 For each frame size previously defined under `Default Test
1973 Parameters <#DefaultParams>`__ and for each of the following number of
1983 - Max supported number of flows.
1985 This test will be conducted under two conditions following the
1986 establishment of all flows as required by RFC 2544, regarding the flow
1987 expiration time-out:
1989 1) The time-out never expires during each trial.
1991 2) The time-out expires for all flows periodically. This would require a
1992 short time-out compared with flow re-appearance for a small number of
1993 flows, and may not be possible for all flow conditions.
1995 The maximum 0% packet loss Throughput should be determined in a manner
1996 identical to LTD.Throughput.RFC2544.PacketLossRatio.
1998 **Expected Result**:
2000 **Metrics Collected**:
2002 The following are the metrics collected for this test:
2004 - The maximum number of frames per second that can be forwarded at the
2005 specified number of flows and the specified frame size, with zero
2010 Test ID: LTD.MemoryBandwidth.RFC2544.0PacketLoss.Scalability
2011 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2012 **Title**: RFC 2544 0% loss Memory Bandwidth Scalability test
2014 **Prerequisite Tests**: LTD.Throughput.RFC2544.PacketLossRatio, IF the
2015 delta Throughput between an undisturbed RFC2544 test and this test with
2016 the Throughput affected by cache and memory bandwidth contention is desired.
2022 The aim of this test is to understand how the DUT's performance is
2023 affected by cache sharing and memory bandwidth between processes.
2025 During the test all cores not used by the vSwitch should be running a
2026 memory intensive application. This application should read and write
2027 random data to random addresses in unused physical memory. The random
2028 nature of the data and addresses is intended to consume cache, exercise
2029 main memory access (as opposed to cache) and exercise all memory buses
2030 equally. Furthermore:
2032 - the ratio of reads to writes should be recorded. A ratio of 1:1
2034 - the reads and writes MUST be of cache-line size and be cache-line aligned.
2035 - in NUMA architectures memory access SHOULD be local to the core's node.
2036 Whether only local memory or a mix of local and remote memory is used
2038 - the memory bandwidth (reads plus writes) used per-core MUST be recorded;
2039 the test MUST be run with a per-core memory bandwidth equal to half the
2040 maximum system memory bandwidth divided by the number of cores. The test
2041 MAY be run with other values for the per-core memory bandwidth.
2042 - the test MAY also be run with the memory intensive application running
2045 Under these conditions the DUT's 0% packet loss throughput is determined
2046 as per LTD.Throughput.RFC2544.PacketLossRatio.
2048 **Expected Result**:
2050 **Metrics Collected**:
2052 The following are the metrics collected for this test:
2054 - The DUT's 0% packet loss throughput in the presence of cache sharing and
2055 memory bandwidth between processes.
2060 -----------------------
2061 The general aim of these tests is to understand the capacity of the
2062 and speed with which the vswitch can accommodate new flows.
2066 Test ID: LTD.Activation.RFC2889.AddressCachingCapacity
2067 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2068 **Title**: RFC2889 Address Caching Capacity Test
2070 **Prerequisite Test**: N/A
2076 Please note this test is only applicable to virtual switches that are capable of
2077 MAC learning. The aim of this test is to determine the address caching
2078 capacity of the DUT for a constant load (fixed length frames at a fixed
2079 interval time). The selected frame sizes are those previously defined
2080 under `Default Test Parameters <#DefaultParams>`__.
2082 In order to run this test the aging time, that is the maximum time the
2083 DUT will keep a learned address in its flow table, and a set of initial
2084 addresses, whose value should be >= 1 and <= the max number supported by
2085 the implementation must be known. Please note that if the aging time is
2086 configurable it must be longer than the time necessary to produce frames
2087 from the external source at the specified rate. If the aging time is
2088 fixed the frame rate must be brought down to a value that the external
2089 source can produce in a time that is less than the aging time.
2091 Learning Frames should be sent from an external source to the DUT to
2092 install a number of flows. The Learning Frames must have a fixed
2093 destination address and must vary the source address of the frames. The
2094 DUT should install flows in its flow table based on the varying source
2095 addresses. Frames should then be transmitted from an external source at
2096 a suitable frame rate to see if the DUT has properly learned all of the
2097 addresses. If there is no frame loss and no flooding, the number of
2098 addresses sent to the DUT should be increased and the test is repeated
2099 until the max number of cached addresses supported by the DUT
2102 **Expected Result**:
2104 **Metrics collected**:
2106 The following are the metrics collected for this test:
2108 - Number of cached addresses supported by the DUT.
2109 - CPU and memory utilization may also be collected as part of this
2110 test, to determine the vSwitch's performance footprint on the system.
2112 **Deployment scenario**:
2114 - Physical → virtual switch → 2 x physical (one receiving, one listening).
2118 Test ID: LTD.Activation.RFC2889.AddressLearningRate
2119 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2120 **Title**: RFC2889 Address Learning Rate Test
2122 **Prerequisite Test**: LTD.Memory.RFC2889.AddressCachingCapacity
2128 Please note this test is only applicable to virtual switches that are capable of
2129 MAC learning. The aim of this test is to determine the rate of address
2130 learning of the DUT for a constant load (fixed length frames at a fixed
2131 interval time). The selected frame sizes are those previously defined
2132 under `Default Test Parameters <#DefaultParams>`__, traffic should be
2133 sent with each IPv4/IPv6 address incremented by one. The rate at which
2134 the DUT learns a new address should be measured. The maximum caching
2135 capacity from LTD.Memory.RFC2889.AddressCachingCapacity should be taken
2136 into consideration as the maximum number of addresses for which the
2137 learning rate can be obtained.
2139 **Expected Result**: It may be worthwhile to report the behaviour when
2140 operating beyond address capacity - some DUTs may be more friendly to
2141 new addresses than others.
2143 **Metrics collected**:
2145 The following are the metrics collected for this test:
2147 - The address learning rate of the DUT.
2149 **Deployment scenario**:
2151 - Physical → virtual switch → 2 x physical (one receiving, one listening).
2155 Coupling between control path and datapath Tests
2156 -------------------------------------------------------
2157 The following tests aim to determine how tightly coupled the datapath
2158 and the control path are within a virtual switch. The following list
2159 is not exhaustive but should indicate the type of tests that should be
2160 required. It is expected that more will be added.
2164 Test ID: LTD.CPDPCouplingFlowAddition
2165 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2166 **Title**: Control Path and Datapath Coupling
2168 **Prerequisite Test**:
2174 The aim of this test is to understand how exercising the DUT's control
2175 path affects datapath performance.
2177 Initially a certain number of flow table entries are installed in the
2178 vSwitch. Then over the duration of an RFC2544 throughput test
2179 flow-entries are added and removed at the rates specified below. No
2180 traffic is 'hitting' these flow-entries, they are simply added and
2183 The test MUST be repeated with the following initial number of
2184 flow-entries installed: - < 10 - 1000 - 100,000 - 10,000,000 (or the
2185 maximum supported number of flow-entries)
2187 The test MUST be repeated with the following rates of flow-entry
2188 addition and deletion per second: - 0 - 1 (i.e. 1 addition plus 1
2189 deletion) - 100 - 10,000
2191 **Expected Result**:
2193 **Metrics Collected**:
2195 The following are the metrics collected for this test:
2197 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
2199 - The average latency of the traffic flow when passing through the DUT
2200 (if testing for latency, note that this average is different from the
2201 test specified in Section 26.3 of
2202 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
2203 - CPU and memory utilization may also be collected as part of this
2204 test, to determine the vSwitch's performance footprint on the system.
2206 **Deployment scenario**:
2208 - Physical → virtual switch → physical.
2212 CPU and memory consumption
2213 ---------------------------------
2214 The following tests will profile a virtual switch's CPU and memory
2215 utilization under various loads and circumstances. The following
2216 list is not exhaustive but should indicate the type of tests that
2217 should be required. It is expected that more will be added.
2221 Test ID: LTD.CPU.RFC2544.0PacketLoss
2222 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2223 **Title**: RFC 2544 0% Loss Compute Test
2225 **Prerequisite Test**:
2231 The aim of this test is to understand the overall performance of the
2232 system when a CPU intensive application is run on the same DUT as the
2233 Virtual Switch. For each frame size, an
2234 LTD.Throughput.RFC2544.PacketLossRatio (0% Packet Loss) test should be
2235 performed. Throughout the entire test a CPU intensive application should
2236 be run on all cores on the system not in use by the Virtual Switch. For
2237 NUMA system only cores on the same NUMA node are loaded.
2239 It is recommended that stress-ng be used for loading the non-Virtual
2240 Switch cores but any stress tool MAY be used.
2242 **Expected Result**:
2244 **Metrics Collected**:
2246 The following are the metrics collected for this test:
2248 - CPU utilization of the cores running the Virtual Switch.
2249 - The number of identity of the cores allocated to the Virtual Switch.
2250 - The configuration of the stress tool (for example the command line
2251 parameters used to start it.)
2255 Summary List of Tests
2256 ----------------------------
2259 - Test ID: LTD.Throughput.RFC2544.PacketLossRatio
2260 - Test ID: LTD.Throughput.RFC2544.PacketLossRatioFrameModification
2261 - Test ID: LTD.Throughput.RFC2544.Profile
2262 - Test ID: LTD.Throughput.RFC2544.SystemRecoveryTime
2263 - Test ID: LTD.Throughput.RFC2544.BackToBackFrames
2264 - Test ID: LTD.Throughput.RFC2889.Soak
2265 - Test ID: LTD.Throughput.RFC2889.SoakFrameModification
2266 - Test ID: LTD.Throughput.RFC6201.ResetTime
2267 - Test ID: LTD.Throughput.RFC2889.MaxForwardingRate
2268 - Test ID: LTD.Throughput.RFC2889.ForwardPressure
2269 - Test ID: LTD.Throughput.RFC2889.ErrorFramesFiltering
2270 - Test ID: LTD.Throughput.RFC2889.BroadcastFrameForwarding
2272 2. Packet Latency tests
2274 - Test ID: LTD.PacketLatency.InitialPacketProcessingLatency
2275 - Test ID: LTD.PacketDelayVariation.RFC3393.Soak
2277 3. Scalability tests
2279 - Test ID: LTD.Scalability.RFC2544.0PacketLoss
2280 - Test ID: LTD.MemoryBandwidth.RFC2544.0PacketLoss.Scalability
2284 - Test ID: LTD.Activation.RFC2889.AddressCachingCapacity
2285 - Test ID: LTD.Activation.RFC2889.AddressLearningRate
2287 5. Coupling between control path and datapath Tests
2289 - Test ID: LTD.CPDPCouplingFlowAddition
2291 6. CPU and memory consumption
2293 - Test ID: LTD.CPU.RFC2544.0PacketLoss
Code Review / vswitchperf.git / blob