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 details-of-LTD_, preceded by the doc-id_ and the scope_.
18 This document is currently in draft form.
26 =========================
28 The document id will be used to uniquely
29 identify versions of the LTD. The format for the document id will be:
30 OPNFV\_vswitchperf\_LTD\_REL\_STATUS, where by the
31 status is one of: draft, reviewed, corrected or final. The document id
32 for this version of the LTD is:
33 OPNFV\_vswitchperf\_LTD\_Brahmaputra\_REVIEWED.
42 The main purpose of this project is to specify a suite of
43 performance tests in order to objectively measure the current packet
44 transfer characteristics of a virtual switch in the NFVI. The intent of
45 the project is to facilitate testing of any virtual switch. Thus, a
46 generic suite of tests shall be developed, with no hard dependencies to
47 a single implementation. In addition, the test case suite shall be
48 architecture independent.
50 The test cases developed in this project shall not form part of a
51 separate test framework, all of these tests may be inserted into the
52 Continuous Integration Test Framework and/or the Platform Functionality
53 Test Framework - if a vSwitch becomes a standard component of an OPNFV
61 * `RFC 1242 Benchmarking Terminology for Network Interconnection
62 Devices <http://www.ietf.org/rfc/rfc1242.txt>`__
63 * `RFC 2544 Benchmarking Methodology for Network Interconnect
64 Devices <http://www.ietf.org/rfc/rfc2544.txt>`__
65 * `RFC 2285 Benchmarking Terminology for LAN Switching
66 Devices <http://www.ietf.org/rfc/rfc2285.txt>`__
67 * `RFC 2889 Benchmarking Methodology for LAN Switching
68 Devices <http://www.ietf.org/rfc/rfc2889.txt>`__
69 * `RFC 3918 Methodology for IP Multicast
70 Benchmarking <http://www.ietf.org/rfc/rfc3918.txt>`__
71 * `RFC 4737 Packet Reordering
72 Metrics <http://www.ietf.org/rfc/rfc4737.txt>`__
73 * `RFC 5481 Packet Delay Variation Applicability
74 Statement <http://www.ietf.org/rfc/rfc5481.txt>`__
75 * `RFC 6201 Device Reset
76 Characterization <http://tools.ietf.org/html/rfc6201>`__
82 ===================================
83 Details of the Level Test Design
84 ===================================
86 This section describes the features to be tested (
87 FeaturesToBeTested_), the test approach (Approach_);
88 it also identifies the sets of test cases or scenarios (
89 TestIdentification_) along with the pass/fail criteria and
90 the test deliverables.
94 .. _FeaturesToBeTested:
97 ==========================
99 Characterizing virtual switches (i.e. Device Under Test (DUT) in this document)
100 includes measuring the following performance metrics:
102 - **Throughput** as defined by `RFC1242
103 <https://www.rfc-editor.org/rfc/rfc1242.txt>`__: The maximum rate at which
104 **none** of the offered frames are dropped by the DUT. The maximum frame
105 rate and bit rate that can be transmitted by the DUT without any error
106 should be recorded. Note there is an equivalent bit rate and a specific
107 layer at which the payloads contribute to the bits. Errors and
108 improperly formed frames or packets are dropped.
109 - **Packet delay** introduced by the DUT and its cumulative effect on
110 E2E networks. Frame delay can be measured equivalently.
111 - **Packet delay variation**: measured from the perspective of the
112 VNF/application. Packet delay variation is sometimes called "jitter".
113 However, we will avoid the term "jitter" as the term holds different
114 meaning to different groups of people. In this document we will
115 simply use the term packet delay variation. The preferred form for this
116 metric is the PDV form of delay variation defined in `RFC5481
117 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__. The most relevant
118 measurement of PDV considers the delay variation of a single user flow,
119 as this will be relevant to the size of end-system buffers to compensate
120 for delay variation. The measurement system's ability to store the
121 delays of individual packets in the flow of interest is a key factor
122 that determines the specific measurement method. At the outset, it is
123 ideal to view the complete PDV distribution. Systems that can capture
124 and store packets and their delays have the freedom to calculate the
125 reference minimum delay and to determine various quantiles of the PDV
126 distribution accurately (in post-measurement processing routines).
127 Systems without storage must apply algorithms to calculate delay and
128 statistical measurements on the fly. For example, a system may store
129 temporary estimates of the mimimum delay and the set of (100) packets
130 with the longest delays during measurement (to calculate a high quantile,
131 and update these sets with new values periodically.
132 In some cases, a limited number of delay histogram bins will be
133 available, and the bin limits will need to be set using results from
134 repeated experiments. See section 8 of `RFC5481
135 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__.
136 - **Packet loss** (within a configured waiting time at the receiver): All
137 packets sent to the DUT should be accounted for.
138 - **Burst behaviour**: measures the ability of the DUT to buffer packets.
139 - **Packet re-ordering**: measures the ability of the device under test to
140 maintain sending order throughout transfer to the destination.
141 - **Packet correctness**: packets or Frames must be well-formed, in that
142 they include all required fields, conform to length requirements, pass
143 integrity checks, etc.
144 - **Availability and capacity** of the DUT i.e. when the DUT is fully “up”
145 and connected, following measurements should be captured for
146 DUT without any network packet load:
148 - Includes average power consumption of the CPUs (in various power states) and
149 system over specified period of time. Time period should not be less
151 - Includes average per core CPU utilization over specified period of time.
152 Time period should not be less than 60 seconds.
153 - Includes the number of NIC interfaces supported.
154 - Includes headroom of VM workload processing cores (i.e. available
164 In order to determine the packet transfer characteristics of a virtual
165 switch, the tests will be broken down into the following categories:
170 ----------------------
171 - **Throughput Tests** to measure the maximum forwarding rate (in
172 frames per second or fps) and bit rate (in Mbps) for a constant load
173 (as defined by `RFC1242 <https://www.rfc-editor.org/rfc/rfc1242.txt>`__)
174 without traffic loss.
175 - **Packet and Frame Delay Tests** to measure average, min and max
176 packet and frame delay for constant loads.
177 - **Stream Performance Tests** (TCP, UDP) to measure bulk data transfer
178 performance, i.e. how fast systems can send and receive data through
180 - **Request/Response Performance** Tests (TCP, UDP) the measure the
181 transaction rate through the virtual switch.
182 - **Packet Delay Tests** to understand latency distribution for
183 different packet sizes and over an extended test run to uncover
185 - **Scalability Tests** to understand how the virtual switch performs
186 as the number of flows, active ports, complexity of the forwarding
187 logic's configuration... it has to deal with increases.
188 - **Control Path and Datapath Coupling** Tests, to understand how
189 closely coupled the datapath and the control path are as well as the
190 effect of this coupling on the performance of the DUT.
191 - **CPU and Memory Consumption Tests** to understand the virtual
192 switch’s footprint on the system, this includes:
194 * CPU core utilization.
195 * CPU cache utilization.
197 * System bus (QPI, PCI, ..) utilization.
198 * Memory lanes utilization.
199 * CPU cycles consumed per packet.
200 * Time To Establish Flows Tests.
202 - **Noisy Neighbour Tests**, to understand the effects of resource
203 sharing on the performance of a virtual switch.
205 **Note:** some of the tests above can be conducted simultaneously where
206 the combined results would be insightful, for example Packet/Frame Delay
212 --------------------------
213 The following represents possible deployment test scenarios which can
214 help to determine the performance of both the virtual switch and the
215 datapaths to physical ports (to NICs) and to logical ports (to VNFs):
219 Physical port → vSwitch → physical port
220 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
221 .. code-block:: console
224 +--------------------------------------------------+ |
225 | +--------------------+ | |
228 | +--------------+ +--------------+ | |
229 | | phy port | vSwitch | phy port | | |
230 +---+--------------+------------+--------------+---+ _|
234 +--------------------------------------------------+
236 | traffic generator |
238 +--------------------------------------------------+
242 Physical port → vSwitch → VNF → vSwitch → physical port
243 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
244 .. code-block:: console
247 +---------------------------------------------------+ |
249 | +-------------------------------------------+ | |
250 | | Application | | |
251 | +-------------------------------------------+ | |
255 | +---------------+ +---------------+ | |
256 | | logical port 0| | logical port 1| | |
257 +---+---------------+-----------+---------------+---+ _|
261 +---+---------------+----------+---------------+---+ |
262 | | logical port 0| | logical port 1| | |
263 | +---------------+ +---------------+ | |
267 | +--------------+ +--------------+ | |
268 | | phy port | vSwitch | phy port | | |
269 +---+--------------+------------+--------------+---+ _|
273 +--------------------------------------------------+
275 | traffic generator |
277 +--------------------------------------------------+
281 Physical port → vSwitch → VNF → vSwitch → VNF → vSwitch → physical port
282 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
284 .. code-block:: console
287 +----------------------+ +----------------------+ |
288 | Guest 1 | | Guest 2 | |
289 | +---------------+ | | +---------------+ | |
290 | | Application | | | | Application | | |
291 | +---------------+ | | +---------------+ | |
293 | | v | | | v | | Guests
294 | +---------------+ | | +---------------+ | |
295 | | logical ports | | | | logical ports | | |
296 | | 0 1 | | | | 0 1 | | |
297 +---+---------------+--+ +---+---------------+--+ _|
301 +---+---------------+---------+---------------+--+ |
302 | | 0 1 | | 3 4 | | |
303 | | logical ports | | logical ports | | |
304 | +---------------+ +---------------+ | |
306 | | L-----------------+ v | |
307 | +--------------+ +--------------+ | |
308 | | phy ports | vSwitch | phy ports | | |
309 +---+--------------+----------+--------------+---+ _|
313 +--------------------------------------------------+
315 | traffic generator |
317 +--------------------------------------------------+
321 Physical port → VNF → vSwitch → VNF → physical port
322 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
324 .. code-block:: console
327 +----------------------+ +----------------------+ |
328 | Guest 1 | | Guest 2 | |
329 |+-------------------+ | | +-------------------+| |
330 || Application | | | | Application || |
331 |+-------------------+ | | +-------------------+| |
332 | ^ | | | ^ | | | Guests
334 |+-------------------+ | | +-------------------+| |
335 || logical ports | | | | logical ports || |
336 || 0 1 | | | | 0 1 || |
337 ++--------------------++ ++--------------------++ _|
339 (PCI passthrough) | | (PCI passthrough)
341 +--------++------------+-+------------++---------+ |
342 | | || 0 | | 1 || | | |
343 | | ||logical port| |logical port|| | | |
344 | | |+------------+ +------------+| | | |
346 | | | L-----------------+ | | | |
348 | | | vSwitch | | | |
349 | | +-----------------------------+ | | |
352 | +--------------+ +--------------+ | |
353 | | phy port/VF | | phy port/VF | | |
354 +-+--------------+--------------+--------------+-+ _|
358 +--------------------------------------------------+
360 | traffic generator |
362 +--------------------------------------------------+
366 Physical port → vSwitch → VNF
367 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
369 .. code-block:: console
372 +---------------------------------------------------+ |
374 | +-------------------------------------------+ | |
375 | | Application | | |
376 | +-------------------------------------------+ | |
380 | +---------------+ | |
381 | | logical port 0| | |
382 +---+---------------+-------------------------------+ _|
386 +---+---------------+------------------------------+ |
387 | | logical port 0| | |
388 | +---------------+ | |
392 | +--------------+ | |
393 | | phy port | vSwitch | |
394 +---+--------------+------------ -------------- ---+ _|
398 +--------------------------------------------------+
400 | traffic generator |
402 +--------------------------------------------------+
406 VNF → vSwitch → physical port
407 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
409 .. code-block:: console
412 +---------------------------------------------------+ |
414 | +-------------------------------------------+ | |
415 | | Application | | |
416 | +-------------------------------------------+ | |
420 | +---------------+ | |
421 | | logical port | | |
422 +-------------------------------+---------------+---+ _|
426 +------------------------------+---------------+---+ |
427 | | logical port | | |
428 | +---------------+ | |
432 | +--------------+ | |
433 | vSwitch | phy port | | |
434 +-------------------------------+--------------+---+ _|
438 +--------------------------------------------------+
440 | traffic generator |
442 +--------------------------------------------------+
446 VNF → vSwitch → VNF → vSwitch
447 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
449 .. code-block:: console
452 +-------------------------+ +-------------------------+ |
453 | Guest 1 | | Guest 2 | |
454 | +-----------------+ | | +-----------------+ | |
455 | | Application | | | | Application | | |
456 | +-----------------+ | | +-----------------+ | |
460 | +---------------+ | | +---------------+ | |
461 | | logical port 0| | | | logical port 0| | |
462 +-----+---------------+---+ +---+---------------+-----+ _|
466 +----+---------------+------------+---------------+-----+ |
467 | | port 0 | | port 1 | | |
468 | +---------------+ +---------------+ | |
471 | +--------------------+ | |
474 +-------------------------------------------------------+ _|
478 HOST 1(Physical port → virtual switch → VNF → virtual switch → Physical port)
479 → HOST 2(Physical port → virtual switch → VNF → virtual switch → Physical port)
481 HOST 1 (PVP) → HOST 2 (PVP)
482 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
484 .. code-block:: console
487 +----------------------+ +----------------------+ |
488 | Guest 1 | | Guest 2 | |
489 | +---------------+ | | +---------------+ | |
490 | | Application | | | | Application | | |
491 | +---------------+ | | +---------------+ | |
493 | | v | | | v | | Guests
494 | +---------------+ | | +---------------+ | |
495 | | logical ports | | | | logical ports | | |
496 | | 0 1 | | | | 0 1 | | |
497 +---+---------------+--+ +---+---------------+--+ _|
501 +---+---------------+--+ +---+---------------+--+ |
502 | | 0 1 | | | | 3 4 | | |
503 | | logical ports | | | | logical ports | | |
504 | +---------------+ | | +---------------+ | |
505 | ^ | | | ^ | | | Hosts
507 | +--------------+ | | +--------------+ | |
508 | | phy ports | | | | phy ports | | |
509 +---+--------------+---+ +---+--------------+---+ _|
511 | +-----------------+ |
513 +--------------------------------------------------+
515 | traffic generator |
517 +--------------------------------------------------+
521 **Note:** For tests where the traffic generator and/or measurement
522 receiver are implemented on VM and connected to the virtual switch
523 through vNIC, the issues of shared resources and interactions between
524 the measurement devices and the device under test must be considered.
526 **Note:** Some RFC 2889 tests require a full-mesh sending and receiving
527 pattern involving more than two ports. This possibility is illustrated in the
528 Physical port → vSwitch → VNF → vSwitch → VNF → vSwitch → physical port
529 diagram above (with 2 sending and 2 receiving ports, though all ports
530 could be used bi-directionally).
532 **Note:** When Deployment Scenarios are used in RFC 2889 address learning
533 or cache capacity testing, an additional port from the vSwitch must be
534 connected to the test device. This port is used to listen for flooded
540 --------------------------
541 To establish the baseline performance of the virtual switch, tests would
542 initially be run with a simple workload in the VNF (the recommended
543 simple workload VNF would be `DPDK <http://www.dpdk.org/>`__'s testpmd
544 application forwarding packets in a VM or vloop\_vnf a simple kernel
545 module that forwards traffic between two network interfaces inside the
546 virtualized environment while bypassing the networking stack).
547 Subsequently, the tests would also be executed with a real Telco
548 workload running in the VNF, which would exercise the virtual switch in
549 the context of higher level Telco NFV use cases, and prove that its
550 underlying characteristics and behaviour can be measured and validated.
551 Suitable real Telco workload VNFs are yet to be identified.
555 Default Test Parameters
556 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
558 The following list identifies the default parameters for suite of
561 - Reference application: Simple forwarding or Open Source VNF.
562 - Frame size (bytes): 64, 128, 256, 512, 1024, 1280, 1518, 2K, 4k OR
563 Packet size based on use-case (e.g. RTP 64B, 256B) OR Mix of packet sizes as
564 maintained by the Functest project <https://wiki.opnfv.org/traffic_profile_management>.
565 - Reordering check: Tests should confirm that packets within a flow are
567 - Duplex: Unidirectional / Bidirectional. Default: Full duplex with
568 traffic transmitting in both directions, as network traffic generally
569 does not flow in a single direction. By default the data rate of
570 transmitted traffic should be the same in both directions, please
571 note that asymmetric traffic (e.g. downlink-heavy) tests will be
572 mentioned explicitly for the relevant test cases.
573 - Number of Flows: Default for non scalability tests is a single flow.
574 For scalability tests the goal is to test with maximum supported
575 flows but where possible will test up to 10 Million flows. Start with
576 a single flow and scale up. By default flows should be added
577 sequentially, tests that add flows simultaneously will explicitly
578 call out their flow addition behaviour. Packets are generated across
579 the flows uniformly with no burstiness. For multi-core tests should
580 consider the number of packet flows based on vSwitch/VNF multi-thread
581 implementation and behavior.
583 - Traffic Types: UDP, SCTP, RTP, GTP and UDP traffic.
584 - Deployment scenarios are:
585 - Physical → virtual switch → physical.
586 - Physical → virtual switch → VNF → virtual switch → physical.
587 - Physical → virtual switch → VNF → virtual switch → VNF → virtual
589 - Physical → VNF → virtual switch → VNF → physical.
590 - Physical → virtual switch → VNF.
591 - VNF → virtual switch → Physical.
592 - VNF → virtual switch → VNF.
594 Tests MUST have these parameters unless otherwise stated. **Test cases
595 with non default parameters will be stated explicitly**.
597 **Note**: For throughput tests unless stated otherwise, test
598 configurations should ensure that traffic traverses the installed flows
599 through the virtual switch, i.e. flows are installed and have an appropriate
600 time out that doesn't expire before packet transmission starts.
605 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
607 Virtual switches classify packets into flows by processing and matching
608 particular header fields in the packet/frame and/or the input port where
609 the packets/frames arrived. The vSwitch then carries out an action on
610 the group of packets that match the classification parameters. Thus a
611 flow is considered to be a sequence of packets that have a shared set of
612 header field values or have arrived on the same port and have the same
613 action applied to them. Performance results can vary based on the
614 parameters the vSwitch uses to match for a flow. The recommended flow
615 classification parameters for L3 vSwitch performance tests are: the
616 input port, the source IP address, the destination IP address and the
617 Ethernet protocol type field. It is essential to increase the flow
618 time-out time on a vSwitch before conducting any performance tests that
619 do not measure the flow set-up time. Normally the first packet of a
620 particular flow will install the flow in the vSwitch which adds an
621 additional latency, subsequent packets of the same flow are not subject
622 to this latency if the flow is already installed on the vSwitch.
627 ~~~~~~~~~~~~~~~~~~~~~
629 Tests will be assigned a priority in order to determine which tests
630 should be implemented immediately and which tests implementations
633 Priority can be of following types: - Urgent: Must be implemented
634 immediately. - High: Must be implemented in the next release. - Medium:
635 May be implemented after the release. - Low: May or may not be
643 The SUT should be configured to its "default" state. The
644 SUT's configuration or set-up must not change between tests in any way
645 other than what is required to do the test. All supported protocols must
646 be configured and enabled for each test set up.
651 ~~~~~~~~~~~~~~~~~~~~~~~~~~
653 The DUT should be configured with n ports where
654 n is a multiple of 2. Half of the ports on the DUT should be used as
655 ingress ports and the other half of the ports on the DUT should be used
656 as egress ports. Where a DUT has more than 2 ports, the ingress data
657 streams should be set-up so that they transmit packets to the egress
658 ports in sequence so that there is an even distribution of traffic
659 across ports. For example, if a DUT has 4 ports 0(ingress), 1(ingress),
660 2(egress) and 3(egress), the traffic stream directed at port 0 should
661 output a packet to port 2 followed by a packet to port 3. The traffic
662 stream directed at port 1 should also output a packet to port 2 followed
663 by a packet to port 3.
668 ~~~~~~~~~~~~~~~~~~~~~
670 **Frame formats Layer 2 (data link layer) protocols**
674 .. code-block:: console
676 +---------------------------+-----------+
677 | Ethernet Header | Payload | Check Sum |
678 +-----------------+---------+-----------+
679 |_________________|_________|___________|
680 14 Bytes 46 - 1500 4 Bytes
684 **Layer 3 (network layer) protocols**
688 .. code-block:: console
690 +-----------------+-----------+---------+-----------+
691 | Ethernet Header | IP Header | Payload | Checksum |
692 +-----------------+-----------+---------+-----------+
693 |_________________|___________|_________|___________|
694 14 Bytes 20 bytes 26 - 1480 4 Bytes
699 .. code-block:: console
701 +-----------------+-----------+---------+-----------+
702 | Ethernet Header | IP Header | Payload | Checksum |
703 +-----------------+-----------+---------+-----------+
704 |_________________|___________|_________|___________|
705 14 Bytes 40 bytes 26 - 1460 4 Bytes
708 **Layer 4 (transport layer) protocols**
714 .. code-block:: console
716 +-----------------+-----------+-----------------+---------+-----------+
717 | Ethernet Header | IP Header | Layer 4 Header | Payload | Checksum |
718 +-----------------+-----------+-----------------+---------+-----------+
719 |_________________|___________|_________________|_________|___________|
720 14 Bytes 40 bytes 20 Bytes 6 - 1460 4 Bytes
724 **Layer 5 (application layer) protocols**
729 .. code-block:: console
731 +-----------------+-----------+-----------------+---------+-----------+
732 | Ethernet Header | IP Header | Layer 4 Header | Payload | Checksum |
733 +-----------------+-----------+-----------------+---------+-----------+
734 |_________________|___________|_________________|_________|___________|
735 14 Bytes 20 bytes 20 Bytes >= 6 Bytes 4 Bytes
740 ~~~~~~~~~~~~~~~~~~~~~~~~~
741 There is a difference between an Ethernet frame,
742 an IP packet, and a UDP datagram. In the seven-layer OSI model of
743 computer networking, packet refers to a data unit at layer 3 (network
744 layer). The correct term for a data unit at layer 2 (data link layer) is
745 a frame, and at layer 4 (transport layer) is a segment or datagram.
747 Important concepts related to 10GbE performance are frame rate and
748 throughput. The MAC bit rate of 10GbE, defined in the IEEE standard 802
749 .3ae, is 10 billion bits per second. Frame rate is based on the bit rate
750 and frame format definitions. Throughput, defined in IETF RFC 1242, is
751 the highest rate at which the system under test can forward the offered
754 The frame rate for 10GbE is determined by a formula that divides the 10
755 billion bits per second by the preamble + frame length + inter-frame
758 The maximum frame rate is calculated using the minimum values of the
759 following parameters, as described in the IEEE 802 .3ae standard:
761 - Preamble: 8 bytes \* 8 = 64 bits
762 - Frame Length: 64 bytes (minimum) \* 8 = 512 bits
763 - Inter-frame Gap: 12 bytes (minimum) \* 8 = 96 bits
765 Therefore, Maximum Frame Rate (64B Frames)
766 = MAC Transmit Bit Rate / (Preamble + Frame Length + Inter-frame Gap)
767 = 10,000,000,000 / (64 + 512 + 96)
768 = 10,000,000,000 / 672
769 = 14,880,952.38 frame per second (fps)
773 System isolation and validation
774 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
776 A key consideration when conducting any sort of benchmark is trying to
777 ensure the consistency and repeatability of test results between runs.
778 When benchmarking the performance of a virtual switch there are many
779 factors that can affect the consistency of results. This section
780 describes these factors and the measures that can be taken to limit
781 their effects. In addition, this section will outline some system tests
782 to validate the platform and the VNF before conducting any vSwitch
785 **System Isolation:**
787 When conducting a benchmarking test on any SUT, it is essential to limit
788 (and if reasonable, eliminate) any noise that may interfere with the
789 accuracy of the metrics collected by the test. This noise may be
790 introduced by other hardware or software (OS, other applications), and
791 can result in significantly varying performance metrics being collected
792 between consecutive runs of the same test. In the case of characterizing
793 the performance of a virtual switch, there are a number of configuration
794 parameters that can help increase the repeatability and stability of
795 test results, including:
797 - OS/GRUB configuration:
799 - maxcpus = n where n >= 0; limits the kernel to using 'n'
800 processors. Only use exactly what you need.
801 - isolcpus: Isolate CPUs from the general scheduler. Isolate all
802 CPUs bar one which will be used by the OS.
803 - use taskset to affinitize the forwarding application and the VNFs
804 onto isolated cores. VNFs and the vSwitch should be allocated
805 their own cores, i.e. must not share the same cores. vCPUs for the
806 VNF should be affinitized to individual cores also.
807 - Limit the amount of background applications that are running and
808 set OS to boot to runlevel 3. Make sure to kill any unnecessary
809 system processes/daemons.
810 - Only enable hardware that you need to use for your test – to
811 ensure there are no other interrupts on the system.
812 - Configure NIC interrupts to only use the cores that are not
813 allocated to any other process (VNF/vSwitch).
815 - NUMA configuration: Any unused sockets in a multi-socket system
817 - CPU pinning: The vSwitch and the VNF should each be affinitized to
818 separate logical cores using a combination of maxcpus, isolcpus and
820 - BIOS configuration: BIOS should be configured for performance where
821 an explicit option exists, sleep states should be disabled, any
822 virtualization optimization technologies should be enabled, and
823 hyperthreading should also be enabled, turbo boost and overclocking
826 **System Validation:**
828 System validation is broken down into two sub-categories: Platform
829 validation and VNF validation. The validation test itself involves
830 verifying the forwarding capability and stability for the sub-system
831 under test. The rationale behind system validation is two fold. Firstly
832 to give a tester confidence in the stability of the platform or VNF that
833 is being tested; and secondly to provide base performance comparison
834 points to understand the overhead introduced by the virtual switch.
836 * Benchmark platform forwarding capability: This is an OPTIONAL test
837 used to verify the platform and measure the base performance (maximum
838 forwarding rate in fps and latency) that can be achieved by the
839 platform without a vSwitch or a VNF. The following diagram outlines
840 the set-up for benchmarking Platform forwarding capability:
842 .. code-block:: console
845 +--------------------------------------------------+ |
846 | +------------------------------------------+ | |
848 | | l2fw or DPDK L2FWD app | | Host
850 | +------------------------------------------+ | |
852 +---+------------------------------------------+---+ __|
856 +--------------------------------------------------+
858 | traffic generator |
860 +--------------------------------------------------+
862 * Benchmark VNF forwarding capability: This test is used to verify
863 the VNF and measure the base performance (maximum forwarding rate in
864 fps and latency) that can be achieved by the VNF without a vSwitch.
865 The performance metrics collected by this test will serve as a key
866 comparison point for NIC passthrough technologies and vSwitches. VNF
867 in this context refers to the hypervisor and the VM. The following
868 diagram outlines the set-up for benchmarking VNF forwarding
871 .. code-block:: console
874 +--------------------------------------------------+ |
875 | +------------------------------------------+ | |
879 | +------------------------------------------+ | |
880 | | Passthrough/SR-IOV | | Host
881 | +------------------------------------------+ | |
883 +---+------------------------------------------+---+ __|
887 +--------------------------------------------------+
889 | traffic generator |
891 +--------------------------------------------------+
894 **Methodology to benchmark Platform/VNF forwarding capability**
897 The recommended methodology for the platform/VNF validation and
898 benchmark is: - Run `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
899 Maximum Forwarding Rate test, this test will produce maximum
900 forwarding rate and latency results that will serve as the
901 expected values. These expected values can be used in
902 subsequent steps or compared with in subsequent validation tests. -
903 Transmit bidirectional traffic at line rate/max forwarding rate
904 (whichever is higher) for at least 72 hours, measure throughput (fps)
905 and latency. - Note: Traffic should be bidirectional. - Establish a
906 baseline forwarding rate for what the platform can achieve. - Additional
907 validation: After the test has completed for 72 hours run bidirectional
908 traffic at the maximum forwarding rate once more to see if the system is
909 still functional and measure throughput (fps) and latency. Compare the
910 measure the new obtained values with the expected values.
912 **NOTE 1**: How the Platform is configured for its forwarding capability
913 test (BIOS settings, GRUB configuration, runlevel...) is how the
914 platform should be configured for every test after this
916 **NOTE 2**: How the VNF is configured for its forwarding capability test
917 (# of vCPUs, vNICs, Memory, affinitization…) is how it should be
918 configured for every test that uses a VNF after this.
922 RFCs for testing virtual switch performance
923 --------------------------------------------------
925 The starting point for defining the suite of tests for benchmarking the
926 performance of a virtual switch is to take existing RFCs and standards
927 that were designed to test their physical counterparts and adapting them
928 for testing virtual switches. The rationale behind this is to establish
929 a fair comparison between the performance of virtual and physical
930 switches. This section outlines the RFCs that are used by this
935 RFC 1242 Benchmarking Terminology for Network Interconnection
936 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
937 Devices RFC 1242 defines the terminology that is used in describing
938 performance benchmarking tests and their results. Definitions and
939 discussions covered include: Back-to-back, bridge, bridge/router,
940 constant load, data link frame size, frame loss rate, inter frame gap,
941 latency, and many more.
945 RFC 2544 Benchmarking Methodology for Network Interconnect Devices
946 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
947 RFC 2544 outlines a benchmarking methodology for network Interconnect
948 Devices. The methodology results in performance metrics such as latency,
949 frame loss percentage, and maximum data throughput.
951 In this document network “throughput” (measured in millions of frames
952 per second) is based on RFC 2544, unless otherwise noted. Frame size
953 refers to Ethernet frames ranging from smallest frames of 64 bytes to
954 largest frames of 9K bytes.
958 1. Throughput test defines the maximum number of frames per second
959 that can be transmitted without any error.
961 2. Latency test measures the time required for a frame to travel from
962 the originating device through the network to the destination device.
963 Please note that RFC2544 Latency measurement will be superseded with
964 a measurement of average latency over all successfully transferred
967 3. Frame loss test measures the network’s
968 response in overload conditions - a critical indicator of the
969 network’s ability to support real-time applications in which a
970 large amount of frame loss will rapidly degrade service quality.
972 4. Burst test assesses the buffering capability of a virtual switch. It
973 measures the maximum number of frames received at full line rate
974 before a frame is lost. In carrier Ethernet networks, this
975 measurement validates the excess information rate (EIR) as defined in
978 5. System recovery to characterize speed of recovery from an overload
981 6. Reset to characterize speed of recovery from device or software
982 reset. This type of test has been updated by `RFC6201
983 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__ as such,
984 the methodology defined by this specification will be that of RFC 6201.
986 Although not included in the defined RFC 2544 standard, another crucial
987 measurement in Ethernet networking is packet delay variation. The
988 definition set out by this specification comes from
989 `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__.
993 RFC 2285 Benchmarking Terminology for LAN Switching Devices
994 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
995 RFC 2285 defines the terminology that is used to describe the
996 terminology for benchmarking a LAN switching device. It extends RFC
997 1242 and defines: DUTs, SUTs, Traffic orientation and distribution,
998 bursts, loads, forwarding rates, etc.
1002 RFC 2889 Benchmarking Methodology for LAN Switching
1003 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1004 RFC 2889 outlines a benchmarking methodology for LAN switching, it
1005 extends RFC 2544. The outlined methodology gathers performance
1006 metrics for forwarding, congestion control, latency, address handling
1007 and finally filtering.
1011 RFC 3918 Methodology for IP Multicast Benchmarking
1012 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1013 RFC 3918 outlines a methodology for IP Multicast benchmarking.
1017 RFC 4737 Packet Reordering Metrics
1018 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1019 RFC 4737 describes metrics for identifying and counting re-ordered
1020 packets within a stream, and metrics to measure the extent each
1021 packet has been re-ordered.
1025 RFC 5481 Packet Delay Variation Applicability Statement
1026 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1027 RFC 5481 defined two common, but different forms of delay variation
1028 metrics, and compares the metrics over a range of networking
1029 circumstances and tasks. The most suitable form for vSwitch
1030 benchmarking is the "PDV" form.
1034 RFC 6201 Device Reset Characterization
1035 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1036 RFC 6201 extends the methodology for characterizing the speed of
1037 recovery of the DUT from device or software reset described in RFC
1042 Details of the Test Report
1043 ---------------------------------
1045 There are a number of parameters related to the system, DUT and tests
1046 that can affect the repeatability of a test results and should be
1047 recorded. In order to minimise the variation in the results of a test,
1048 it is recommended that the test report includes the following information:
1050 - Hardware details including:
1053 - Processor details.
1054 - Memory information (see below)
1055 - Number of enabled cores.
1056 - Number of cores used for the test.
1057 - Number of physical NICs, as well as their details (manufacturer,
1058 versions, type and the PCI slot they are plugged into).
1059 - NIC interrupt configuration.
1060 - BIOS version, release date and any configurations that were
1063 - Software details including:
1065 - OS version (for host and VNF)
1066 - Kernel version (for host and VNF)
1067 - GRUB boot parameters (for host and VNF).
1068 - Hypervisor details (Type and version).
1069 - Selected vSwitch, version number or commit id used.
1070 - vSwitch launch command line if it has been parameterised.
1071 - Memory allocation to the vSwitch – which NUMA node it is using,
1072 and how many memory channels.
1073 - Where the vswitch is built from source: compiler details including
1074 versions and the flags that were used to compile the vSwitch.
1075 - DPDK or any other SW dependency version number or commit id used.
1076 - Memory allocation to a VM - if it's from Hugpages/elsewhere.
1077 - VM storage type: snapshot/independent persistent/independent
1080 - Number of Virtual NICs (vNICs), versions, type and driver.
1081 - Number of virtual CPUs and their core affinity on the host.
1082 - Number vNIC interrupt configuration.
1083 - Thread affinitization for the applications (including the vSwitch
1084 itself) on the host.
1085 - Details of Resource isolation, such as CPUs designated for
1086 Host/Kernel (isolcpu) and CPUs designated for specific processes
1105 - Traffic Information:
1107 - Traffic type - UDP, TCP, IMIX / Other.
1110 - Deployment Scenario.
1112 **Note**: Tests that require additional parameters to be recorded will
1113 explicitly specify this.
1115 .. _TestIdentification:
1120 =========================
1125 ----------------------
1126 The following tests aim to determine the maximum forwarding rate that
1127 can be achieved with a virtual switch. The list is not exhaustive but
1128 should indicate the type of tests that should be required. It is
1129 expected that more will be added.
1133 Test ID: LTD.Throughput.RFC2544.PacketLossRatio
1134 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1135 **Title**: RFC 2544 X% packet loss ratio Throughput and Latency Test
1137 **Prerequisite Test**: N/A
1143 This test determines the DUT's maximum forwarding rate with X% traffic
1144 loss for a constant load (fixed length frames at a fixed interval time).
1145 The default loss percentages to be tested are: - X = 0% - X = 10^-7%
1147 Note: Other values can be tested if required by the user.
1149 The selected frame sizes are those previously defined under `Default
1150 Test Parameters <#DefaultParams>`__. The test can also be used to
1151 determine the average latency of the traffic.
1153 Under the `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__
1154 test methodology, the test duration will
1155 include a number of trials; each trial should run for a minimum period
1156 of 60 seconds. A binary search methodology must be applied for each
1157 trial to obtain the final result.
1159 **Expected Result**: At the end of each trial, the presence or absence
1160 of loss determines the modification of offered load for the next trial,
1161 converging on a maximum rate, or
1162 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ Throughput with X%
1164 The Throughput load is re-used in related
1165 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ tests and other
1168 **Metrics Collected**:
1170 The following are the metrics collected for this test:
1172 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
1173 the DUT for each frame size with X% packet loss.
1174 - The average latency of the traffic flow when passing through the DUT
1175 (if testing for latency, note that this average is different from the
1176 test specified in Section 26.3 of
1177 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
1178 - CPU and memory utilization may also be collected as part of this
1179 test, to determine the vSwitch's performance footprint on the system.
1183 Test ID: LTD.Throughput.RFC2544.PacketLossRatioFrameModification
1184 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1185 **Title**: RFC 2544 X% packet loss Throughput and Latency Test with
1188 **Prerequisite Test**: N/A
1194 This test determines the DUT's maximum forwarding rate with X% traffic
1195 loss for a constant load (fixed length frames at a fixed interval time).
1196 The default loss percentages to be tested are: - X = 0% - X = 10^-7%
1198 Note: Other values can be tested if required by the user.
1200 The selected frame sizes are those previously defined under `Default
1201 Test Parameters <#DefaultParams>`__. The test can also be used to
1202 determine the average latency of the traffic.
1204 Under the `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__
1205 test methodology, the test duration will
1206 include a number of trials; each trial should run for a minimum period
1207 of 60 seconds. A binary search methodology must be applied for each
1208 trial to obtain the final result.
1210 During this test, the DUT must perform the following operations on the
1213 - Perform packet parsing on the DUT's ingress port.
1214 - Perform any relevant address look-ups on the DUT's ingress ports.
1215 - Modify the packet header before forwarding the packet to the DUT's
1216 egress port. Packet modifications include:
1218 - Modifying the Ethernet source or destination MAC address.
1219 - Modifying/adding a VLAN tag. (**Recommended**).
1220 - Modifying/adding a MPLS tag.
1221 - Modifying the source or destination ip address.
1222 - Modifying the TOS/DSCP field.
1223 - Modifying the source or destination ports for UDP/TCP/SCTP.
1224 - Modifying the TTL.
1226 **Expected Result**: The Packet parsing/modifications require some
1227 additional degree of processing resource, therefore the
1228 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__
1229 Throughput is expected to be somewhat lower than the Throughput level
1230 measured without additional steps. The reduction is expected to be
1231 greatest on tests with the smallest packet sizes (greatest header
1234 **Metrics Collected**:
1236 The following are the metrics collected for this test:
1238 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
1239 the DUT for each frame size with X% packet loss and packet
1240 modification operations being performed by the DUT.
1241 - The average latency of the traffic flow when passing through the DUT
1242 (if testing for latency, note that this average is different from the
1243 test specified in Section 26.3 of
1244 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
1245 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1246 PDV form of delay variation on the traffic flow,
1247 using the 99th percentile.
1248 - CPU and memory utilization may also be collected as part of this
1249 test, to determine the vSwitch's performance footprint on the system.
1253 Test ID: LTD.Throughput.RFC2544.Profile
1254 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1255 **Title**: RFC 2544 Throughput and Latency Profile
1257 **Prerequisite Test**: N/A
1263 This test reveals how throughput and latency degrades as the offered
1264 rate varies in the region of the DUT's maximum forwarding rate as
1265 determined by LTD.Throughput.RFC2544.PacketLossRatio (0% Packet Loss).
1266 For example it can be used to determine if the degradation of throughput
1267 and latency as the offered rate increases is slow and graceful or sudden
1270 The selected frame sizes are those previously defined under `Default
1271 Test Parameters <#DefaultParams>`__.
1273 The offered traffic rate is described as a percentage delta with respect
1274 to the DUT's RFC 2544 Throughput as determined by
1275 LTD.Throughput.RFC2544.PacketLoss Ratio (0% Packet Loss case). A delta
1276 of 0% is equivalent to an offered traffic rate equal to the RFC 2544
1277 Throughput; A delta of +50% indicates an offered rate half-way
1278 between the Throughput and line-rate, whereas a delta of
1279 -50% indicates an offered rate of half the maximum rate. Therefore the
1280 range of the delta figure is natuarlly bounded at -100% (zero offered
1281 traffic) and +100% (traffic offered at line rate).
1283 The following deltas to the maximum forwarding rate should be applied:
1285 - -50%, -10%, 0%, +10% & +50%
1287 **Expected Result**: For each packet size a profile should be produced
1288 of how throughput and latency vary with offered rate.
1290 **Metrics Collected**:
1292 The following are the metrics collected for this test:
1294 - The forwarding rate in Frames Per Second (FPS) and Mbps of the DUT
1295 for each delta to the maximum forwarding rate and for each frame
1297 - The average latency for each delta to the maximum forwarding rate and
1298 for each frame size.
1299 - CPU and memory utilization may also be collected as part of this
1300 test, to determine the vSwitch's performance footprint on the system.
1301 - Any failures experienced (for example if the vSwitch crashes, stops
1302 processing packets, restarts or becomes unresponsive to commands)
1303 when the offered load is above Maximum Throughput MUST be recorded
1304 and reported with the results.
1308 Test ID: LTD.Throughput.RFC2544.SystemRecoveryTime
1309 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1310 **Title**: RFC 2544 System Recovery Time Test
1312 **Prerequisite Test** LTD.Throughput.RFC2544.PacketLossRatio
1318 The aim of this test is to determine the length of time it takes the DUT
1319 to recover from an overload condition for a constant load (fixed length
1320 frames at a fixed interval time). The selected frame sizes are those
1321 previously defined under `Default Test Parameters <#DefaultParams>`__,
1322 traffic should be sent to the DUT under normal conditions. During the
1323 duration of the test and while the traffic flows are passing though the
1324 DUT, at least one situation leading to an overload condition for the DUT
1325 should occur. The time from the end of the overload condition to when
1326 the DUT returns to normal operations should be measured to determine
1327 recovery time. Prior to overloading the DUT, one should record the
1328 average latency for 10,000 packets forwarded through the DUT.
1330 The overload condition SHOULD be to transmit traffic at a very high
1331 frame rate to the DUT (150% of the maximum 0% packet loss rate as
1332 determined by LTD.Throughput.RFC2544.PacketLossRatio or line-rate
1333 whichever is lower), for at least 60 seconds, then reduce the frame rate
1334 to 75% of the maximum 0% packet loss rate. A number of time-stamps
1335 should be recorded: - Record the time-stamp at which the frame rate was
1336 reduced and record a second time-stamp at the time of the last frame
1337 lost. The recovery time is the difference between the two timestamps. -
1338 Record the average latency for 10,000 frames after the last frame loss
1339 and continue to record average latency measurements for every 10,000
1340 frames, when latency returns to within 10% of pre-overload levels record
1343 **Expected Result**:
1345 **Metrics collected**
1347 The following are the metrics collected for this test:
1349 - The length of time it takes the DUT to recover from an overload
1351 - The length of time it takes the DUT to recover the average latency to
1352 pre-overload conditions.
1354 **Deployment scenario**:
1356 - Physical → virtual switch → physical.
1360 Test ID: LTD.Throughput.RFC2544.BackToBackFrames
1361 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1362 **Title**: RFC2544 Back To Back Frames Test
1364 **Prerequisite Test**: N
1370 The aim of this test is to characterize the ability of the DUT to
1371 process back-to-back frames. For each frame size previously defined
1372 under `Default Test Parameters <#DefaultParams>`__, a burst of traffic
1373 is sent to the DUT with the minimum inter-frame gap between each frame.
1374 If the number of received frames equals the number of frames that were
1375 transmitted, the burst size should be increased and traffic is sent to
1376 the DUT again. The value measured is the back-to-back value, that is the
1377 maximum burst size the DUT can handle without any frame loss. Please note
1378 a trial must run for a minimum of 2 seconds and should be repeated 50
1379 times (at a minimum).
1381 **Expected Result**:
1383 Tests of back-to-back frames with physical devices have produced
1384 unstable results in some cases. All tests should be repeated in multiple
1385 test sessions and results stability should be examined.
1387 **Metrics collected**
1389 The following are the metrics collected for this test:
1391 - The average back-to-back value across the trials, which is
1392 the number of frames in the longest burst that the DUT will
1393 handle without the loss of any frames.
1394 - CPU and memory utilization may also be collected as part of this
1395 test, to determine the vSwitch's performance footprint on the system.
1397 **Deployment scenario**:
1399 - Physical → virtual switch → physical.
1403 Test ID: LTD.Throughput.RFC2889.MaxForwardingRateSoak
1404 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1405 **Title**: RFC 2889 X% packet loss Max Forwarding Rate Soak Test
1407 **Prerequisite Test** LTD.Throughput.RFC2544.PacketLossRatio
1413 The aim of this test is to understand the Max Forwarding Rate stability
1414 over an extended test duration in order to uncover any outliers. To allow
1415 for an extended test duration, the test should ideally run for 24 hours
1416 or, if this is not possible, for at least 6 hours. For this test, each frame
1417 size must be sent at the highest Throughput rate with X% packet loss, as
1418 determined in the prerequisite test. The default loss percentages to be
1419 tested are: - X = 0% - X = 10^-7%
1421 Note: Other values can be tested if required by the user.
1423 **Expected Result**:
1425 **Metrics Collected**:
1427 The following are the metrics collected for this test:
1429 - Max Forwarding Rate stability of the DUT.
1431 - This means reporting the number of packets lost per time interval
1432 and reporting any time intervals with packet loss. The
1433 `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
1434 Forwarding Rate shall be measured in each interval.
1435 An interval of 60s is suggested.
1437 - CPU and memory utilization may also be collected as part of this
1438 test, to determine the vSwitch's performance footprint on the system.
1439 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1440 PDV form of delay variation on the traffic flow,
1441 using the 99th percentile.
1445 Test ID: LTD.Throughput.RFC2889.MaxForwardingRateSoakFrameModification
1446 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1447 **Title**: RFC 2889 Max Forwarding Rate Soak Test with Frame Modification
1449 **Prerequisite Test**:
1450 LTD.Throughput.RFC2544.PacketLossRatioFrameModification (0% Packet Loss)
1456 The aim of this test is to understand the Max Forwarding Rate stability over an
1457 extended test duration in order to uncover any outliers. To allow for an
1458 extended test duration, the test should ideally run for 24 hours or, if
1459 this is not possible, for at least 6 hour. For this test, each frame
1460 size must be sent at the highest Throughput rate with 0% packet loss, as
1461 determined in the prerequisite test.
1463 During this test, the DUT must perform the following operations on the
1466 - Perform packet parsing on the DUT's ingress port.
1467 - Perform any relevant address look-ups on the DUT's ingress ports.
1468 - Modify the packet header before forwarding the packet to the DUT's
1469 egress port. Packet modifications include:
1471 - Modifying the Ethernet source or destination MAC address.
1472 - Modifying/adding a VLAN tag (**Recommended**).
1473 - Modifying/adding a MPLS tag.
1474 - Modifying the source or destination ip address.
1475 - Modifying the TOS/DSCP field.
1476 - Modifying the source or destination ports for UDP/TCP/SCTP.
1477 - Modifying the TTL.
1479 **Expected Result**:
1481 **Metrics Collected**:
1483 The following are the metrics collected for this test:
1485 - Max Forwarding Rate stability of the DUT.
1487 - This means reporting the number of packets lost per time interval
1488 and reporting any time intervals with packet loss. The
1489 `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
1490 Forwarding Rate shall be measured in each interval.
1491 An interval of 60s is suggested.
1493 - CPU and memory utilization may also be collected as part of this
1494 test, to determine the vSwitch's performance footprint on the system.
1495 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1496 PDV form of delay variation on the traffic flow, using the 99th
1501 Test ID: LTD.Throughput.RFC6201.ResetTime
1502 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1503 **Title**: RFC 6201 Reset Time Test
1505 **Prerequisite Test**: N/A
1511 The aim of this test is to determine the length of time it takes the DUT
1512 to recover from a reset.
1514 Two reset methods are defined - planned and unplanned. A planned reset
1515 requires stopping and restarting the virtual switch by the usual
1516 'graceful' method defined by it's documentation. An unplanned reset
1517 requires simulating a fatal internal fault in the virtual switch - for
1518 example by using kill -SIGKILL on a Linux environment.
1520 Both reset methods SHOULD be exercised.
1522 For each frame size previously defined under `Default Test
1523 Parameters <#DefaultParams>`__, traffic should be sent to the DUT under
1524 normal conditions. During the duration of the test and while the traffic
1525 flows are passing through the DUT, the DUT should be reset and the Reset
1526 time measured. The Reset time is the total time that a device is
1527 determined to be out of operation and includes the time to perform the
1528 reset and the time to recover from it (cf. `RFC6201
1529 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__).
1531 `RFC6201 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__ defines two methods
1532 to measure the Reset time:
1534 - Frame-Loss Method: which requires the monitoring of the number of
1535 lost frames and calculates the Reset time based on the number of
1536 frames lost and the offered rate according to the following
1539 .. code-block:: console
1541 Frames_lost (packets)
1542 Reset_time = -------------------------------------
1543 Offered_rate (packets per second)
1545 - Timestamp Method: which measures the time from which the last frame
1546 is forwarded from the DUT to the time the first frame is forwarded
1547 after the reset. This involves time-stamping all transmitted frames
1548 and recording the timestamp of the last frame that was received prior
1549 to the reset and also measuring the timestamp of the first frame that
1550 is received after the reset. The Reset time is the difference between
1551 these two timestamps.
1553 According to `RFC6201 <https://www.rfc-editor.org/rfc/rfc6201.txt>`__ the
1554 choice of method depends on the test tool's capability; the Frame-Loss
1555 method SHOULD be used if the test tool supports:
1557 * Counting the number of lost frames per stream.
1558 * Transmitting test frame despite the physical link status.
1560 whereas the Timestamp method SHOULD be used if the test tool supports:
1561 * Timestamping each frame.
1562 * Monitoring received frame's timestamp.
1563 * Transmitting frames only if the physical link status is up.
1565 **Expected Result**:
1567 **Metrics collected**
1569 The following are the metrics collected for this test:
1571 * Average Reset Time over the number of trials performed.
1573 Results of this test should include the following information:
1575 * The reset method used.
1576 * Throughput in Fps and Mbps.
1577 * Average Frame Loss over the number of trials performed.
1578 * Average Reset Time in milliseconds over the number of trials performed.
1579 * Number of trials performed.
1580 * Protocol: IPv4, IPv6, MPLS, etc.
1581 * Frame Size in Octets
1582 * Port Media: Ethernet, Gigabit Ethernet (GbE), etc.
1583 * Port Speed: 10 Gbps, 40 Gbps etc.
1584 * Interface Encapsulation: Ethernet, Ethernet VLAN, etc.
1586 **Deployment scenario**:
1588 * Physical → virtual switch → physical.
1592 Test ID: LTD.Throughput.RFC2889.MaxForwardingRate
1593 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1594 **Title**: RFC2889 Forwarding Rate Test
1596 **Prerequisite Test**: LTD.Throughput.RFC2544.PacketLossRatio
1602 This test measures the DUT's Max Forwarding Rate when the Offered Load
1603 is varied between the throughput and the Maximum Offered Load for fixed
1604 length frames at a fixed time interval. The selected frame sizes are
1605 those previously defined under `Default Test
1606 Parameters <#DefaultParams>`__. The throughput is the maximum offered
1607 load with 0% frame loss (measured by the prerequisite test), and the
1608 Maximum Offered Load (as defined by
1609 `RFC2285 <https://www.rfc-editor.org/rfc/rfc2285.txt>`__) is *"the highest
1610 number of frames per second that an external source can transmit to a
1611 DUT/SUT for forwarding to a specified output interface or interfaces"*.
1613 Traffic should be sent to the DUT at a particular rate (TX rate)
1614 starting with TX rate equal to the throughput rate. The rate of
1615 successfully received frames at the destination counted (in FPS). If the
1616 RX rate is equal to the TX rate, the TX rate should be increased by a
1617 fixed step size and the RX rate measured again until the Max Forwarding
1620 The trial duration for each iteration should last for the period of time
1621 needed for the system to reach steady state for the frame size being
1622 tested. Under `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__
1623 (Sec. 5.6.3.1) test methodology, the test
1624 duration should run for a minimum period of 30 seconds, regardless
1625 whether the system reaches steady state before the minimum duration
1628 **Expected Result**: According to
1629 `RFC2889 <https://www.rfc-editor.org/rfc/rfc2289.txt>`__ The Max Forwarding
1630 Rate is the highest forwarding rate of a DUT taken from an iterative set of
1631 forwarding rate measurements. The iterative set of forwarding rate measurements
1632 are made by setting the intended load transmitted from an external source and
1633 measuring the offered load (i.e what the DUT is capable of forwarding). If the
1634 Throughput == the Maximum Offered Load, it follows that Max Forwarding Rate is
1635 equal to the Maximum Offered Load.
1637 **Metrics Collected**:
1639 The following are the metrics collected for this test:
1641 - The Max Forwarding Rate for the DUT for each packet size.
1642 - CPU and memory utilization may also be collected as part of this
1643 test, to determine the vSwitch's performance footprint on the system.
1645 **Deployment scenario**:
1647 - Physical → virtual switch → physical. Note: Full mesh tests with
1648 multiple ingress and egress ports are a key aspect of RFC 2889
1649 benchmarks, and scenarios with both 2 and 4 ports should be tested.
1650 In any case, the number of ports used must be reported.
1654 Test ID: LTD.Throughput.RFC2889.ForwardPressure
1655 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1656 **Title**: RFC2889 Forward Pressure Test
1658 **Prerequisite Test**: LTD.Throughput.RFC2889.MaxForwardingRate
1664 The aim of this test is to determine if the DUT transmits frames with an
1665 inter-frame gap that is less than 12 bytes. This test overloads the DUT
1666 and measures the output for forward pressure. Traffic should be
1667 transmitted to the DUT with an inter-frame gap of 11 bytes, this will
1668 overload the DUT by 1 byte per frame. The forwarding rate of the DUT
1671 **Expected Result**: The forwarding rate should not exceed the maximum
1672 forwarding rate of the DUT collected by
1673 LTD.Throughput.RFC2889.MaxForwardingRate.
1675 **Metrics collected**
1677 The following are the metrics collected for this test:
1679 - Forwarding rate of the DUT in FPS or Mbps.
1680 - CPU and memory utilization may also be collected as part of this
1681 test, to determine the vSwitch's performance footprint on the system.
1683 **Deployment scenario**:
1685 - Physical → virtual switch → physical.
1689 Test ID: LTD.Throughput.RFC2889.ErrorFramesFiltering
1690 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1691 **Title**: RFC2889 Error Frames Filtering Test
1693 **Prerequisite Test**: N/A
1699 The aim of this test is to determine whether the DUT will propagate any
1700 erroneous frames it receives or whether it is capable of filtering out
1701 the erroneous frames. Traffic should be sent with erroneous frames
1702 included within the flow at random intervals. Illegal frames that must
1703 be tested include: - Oversize Frames. - Undersize Frames. - CRC Errored
1704 Frames. - Dribble Bit Errored Frames - Alignment Errored Frames
1706 The traffic flow exiting the DUT should be recorded and checked to
1707 determine if the erroneous frames where passed through the DUT.
1709 **Expected Result**: Broken frames are not passed!
1711 **Metrics collected**
1713 No Metrics are collected in this test, instead it determines:
1715 - Whether the DUT will propagate erroneous frames.
1716 - Or whether the DUT will correctly filter out any erroneous frames
1717 from traffic flow with out removing correct frames.
1719 **Deployment scenario**:
1721 - Physical → virtual switch → physical.
1725 Test ID: LTD.Throughput.RFC2889.BroadcastFrameForwarding
1726 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1727 **Title**: RFC2889 Broadcast Frame Forwarding Test
1729 **Prerequisite Test**: N
1735 The aim of this test is to determine the maximum forwarding rate of the
1736 DUT when forwarding broadcast traffic. For each frame previously defined
1737 under `Default Test Parameters <#DefaultParams>`__, the traffic should
1738 be set up as broadcast traffic. The traffic throughput of the DUT should
1741 The test should be conducted with at least 4 physical ports on the DUT.
1742 The number of ports used MUST be recorded.
1744 As broadcast involves forwarding a single incoming packet to several
1745 destinations, the latency of a single packet is defined as the average
1746 of the latencies for each of the broadcast destinations.
1748 The incoming packet is transmitted on each of the other physical ports,
1749 it is not transmitted on the port on which it was received. The test MAY
1750 be conducted using different broadcasting ports to uncover any
1751 performance differences.
1753 **Expected Result**:
1755 **Metrics collected**:
1757 The following are the metrics collected for this test:
1759 - The forwarding rate of the DUT when forwarding broadcast traffic.
1760 - The minimum, average & maximum packets latencies observed.
1762 **Deployment scenario**:
1764 - Physical → virtual switch 3x physical. In the Broadcast rate testing,
1765 four test ports are required. One of the ports is connected to the test
1766 device, so it can send broadcast frames and listen for miss-routed frames.
1770 Test ID: LTD.Throughput.RFC2544.WorstN-BestN
1771 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1772 **Title**: Modified RFC 2544 X% packet loss ratio Throughput and Latency Test
1774 **Prerequisite Test**: N/A
1780 This test determines the DUT's maximum forwarding rate with X% traffic
1781 loss for a constant load (fixed length frames at a fixed interval time).
1782 The default loss percentages to be tested are: X = 0%, X = 10^-7%
1784 Modified RFC 2544 throughput benchmarking methodology aims to quantify
1785 the throughput measurement variations observed during standard RFC 2544
1786 benchmarking measurements of virtual switches and VNFs. The RFC2544
1787 binary search algorithm is modified to use more samples per test trial
1788 to drive the binary search and yield statistically more meaningful
1789 results. This keeps the heart of the RFC2544 methodology, still relying
1790 on the binary search of throughput at specified loss tolerance, while
1791 providing more useful information about the range of results seen in
1792 testing. Instead of using a single traffic trial per iteration step,
1793 each traffic trial is repeated N times and the success/failure of the
1794 iteration step is based on these N traffic trials. Two types of revised
1795 tests are defined - *Worst-of-N* and *Best-of-N*.
1799 *Worst-of-N* indicates the lowest expected maximum throughput for (
1800 packet size, loss tolerance) when repeating the test.
1802 1. Repeat the same test run N times at a set packet rate, record each
1804 2. Take the WORST result (highest packet loss) out of N result samples,
1805 called the Worst-of-N sample.
1806 3. If Worst-of-N sample has loss less than the set loss tolerance, then
1807 the step is successful - increase the test traffic rate.
1808 4. If Worst-of-N sample has loss greater than the set loss tolerance
1809 then the step failed - decrease the test traffic rate.
1814 *Best-of-N* indicates the highest expected maximum throughput for (
1815 packet size, loss tolerance) when repeating the test.
1817 1. Repeat the same traffic run N times at a set packet rate, record
1819 2. Take the BEST result (least packet loss) out of N result samples,
1820 called the Best-of-N sample.
1821 3. If Best-of-N sample has loss less than the set loss tolerance, then
1822 the step is successful - increase the test traffic rate.
1823 4. If Best-of-N sample has loss greater than the set loss tolerance,
1824 then the step failed - decrease the test traffic rate.
1827 Performing both Worst-of-N and Best-of-N benchmark tests yields lower
1828 and upper bounds of expected maximum throughput under the operating
1829 conditions, giving a very good indication to the user of the
1830 deterministic performance range for the tested setup.
1832 **Expected Result**: At the end of each trial series, the presence or
1833 absence of loss determines the modification of offered load for the
1834 next trial series, converging on a maximum rate, or
1835 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ Throughput
1837 The Throughput load is re-used in related
1838 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__ tests and other
1841 **Metrics Collected**:
1843 The following are the metrics collected for this test:
1845 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
1846 the DUT for each frame size with X% packet loss.
1847 - The average latency of the traffic flow when passing through the DUT
1848 (if testing for latency, note that this average is different from the
1849 test specified in Section 26.3 of
1850 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
1851 - Following may also be collected as part of this test, to determine
1852 the vSwitch's performance footprint on the system:
1853 - CPU core utilization.
1854 - CPU cache utilization.
1856 - System bus (QPI, PCI, ...) utilization.
1857 - CPU cycles consumed per packet.
1861 Packet Latency tests
1862 ---------------------------
1863 These tests will measure the store and forward latency as well as the packet
1864 delay variation for various packet types through the virtual switch. The
1865 following list is not exhaustive but should indicate the type of tests
1866 that should be required. It is expected that more will be added.
1870 Test ID: LTD.PacketLatency.InitialPacketProcessingLatency
1871 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1872 **Title**: Initial Packet Processing Latency
1874 **Prerequisite Test**: N/A
1880 In some virtual switch architectures, the first packets of a flow will
1881 take the system longer to process than subsequent packets in the flow.
1882 This test determines the latency for these packets. The test will
1883 measure the latency of the packets as they are processed by the
1884 flow-setup-path of the DUT. There are two methods for this test, a
1885 recommended method and a nalternative method that can be used if it is
1886 possible to disable the fastpath of the virtual switch.
1888 Recommended method: This test will send 64,000 packets to the DUT, each
1889 belonging to a different flow. Average packet latency will be determined
1890 over the 64,000 packets.
1892 Alternative method: This test will send a single packet to the DUT after
1893 a fixed interval of time. The time interval will be equivalent to the
1894 amount of time it takes for a flow to time out in the virtual switch
1895 plus 10%. Average packet latency will be determined over 1,000,000
1898 This test is intended only for non-learning virtual switches; For learning
1899 virtual switches use RFC2889.
1901 For this test, only unidirectional traffic is required.
1903 **Expected Result**: The average latency for the initial packet of all
1904 flows should be greater than the latency of subsequent traffic.
1906 **Metrics Collected**:
1908 The following are the metrics collected for this test:
1910 - Average latency of the initial packets of all flows that are
1911 processed by the DUT.
1913 **Deployment scenario**:
1915 - Physical → Virtual Switch → Physical.
1919 Test ID: LTD.PacketDelayVariation.RFC3393.Soak
1920 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1921 **Title**: Packet Delay Variation Soak Test
1923 **Prerequisite Tests**: LTD.Throughput.RFC2544.PacketLossRatio (0% Packet Loss)
1929 The aim of this test is to understand the distribution of packet delay
1930 variation for different frame sizes over an extended test duration and
1931 to determine if there are any outliers. To allow for an extended test
1932 duration, the test should ideally run for 24 hours or, if this is not
1933 possible, for at least 6 hour. For this test, each frame size must be
1934 sent at the highest possible throughput with 0% packet loss, as
1935 determined in the prerequisite test.
1937 **Expected Result**:
1939 **Metrics Collected**:
1941 The following are the metrics collected for this test:
1943 - The packet delay variation value for traffic passing through the DUT.
1944 - The `RFC5481 <https://www.rfc-editor.org/rfc/rfc5481.txt>`__
1945 PDV form of delay variation on the traffic flow,
1946 using the 99th percentile, for each 60s interval during the test.
1947 - CPU and memory utilization may also be collected as part of this
1948 test, to determine the vSwitch's performance footprint on the system.
1953 ------------------------
1954 The general aim of these tests is to understand the impact of large flow
1955 table size and flow lookups on throughput. The following list is not
1956 exhaustive but should indicate the type of tests that should be required.
1957 It is expected that more will be added.
1961 Test ID: LTD.Scalability.RFC2544.0PacketLoss
1962 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1963 **Title**: RFC 2544 0% loss Scalability throughput test
1965 **Prerequisite Test**: LTD.Throughput.RFC2544.PacketLossRatio, IF the
1966 delta Throughput between the single-flow RFC2544 test and this test with
1967 a variable number of flows is desired.
1973 The aim of this test is to measure how throughput changes as the number
1974 of flows in the DUT increases. The test will measure the throughput
1975 through the fastpath, as such the flows need to be installed on the DUT
1976 before passing traffic.
1978 For each frame size previously defined under `Default Test
1979 Parameters <#DefaultParams>`__ and for each of the following number of
1989 - Max supported number of flows.
1991 This test will be conducted under two conditions following the
1992 establishment of all flows as required by RFC 2544, regarding the flow
1993 expiration time-out:
1995 1) The time-out never expires during each trial.
1997 2) The time-out expires for all flows periodically. This would require a
1998 short time-out compared with flow re-appearance for a small number of
1999 flows, and may not be possible for all flow conditions.
2001 The maximum 0% packet loss Throughput should be determined in a manner
2002 identical to LTD.Throughput.RFC2544.PacketLossRatio.
2004 **Expected Result**:
2006 **Metrics Collected**:
2008 The following are the metrics collected for this test:
2010 - The maximum number of frames per second that can be forwarded at the
2011 specified number of flows and the specified frame size, with zero
2016 Test ID: LTD.MemoryBandwidth.RFC2544.0PacketLoss.Scalability
2017 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2018 **Title**: RFC 2544 0% loss Memory Bandwidth Scalability test
2020 **Prerequisite Tests**: LTD.Throughput.RFC2544.PacketLossRatio, IF the
2021 delta Throughput between an undisturbed RFC2544 test and this test with
2022 the Throughput affected by cache and memory bandwidth contention is desired.
2028 The aim of this test is to understand how the DUT's performance is
2029 affected by cache sharing and memory bandwidth between processes.
2031 During the test all cores not used by the vSwitch should be running a
2032 memory intensive application. This application should read and write
2033 random data to random addresses in unused physical memory. The random
2034 nature of the data and addresses is intended to consume cache, exercise
2035 main memory access (as opposed to cache) and exercise all memory buses
2036 equally. Furthermore:
2038 - the ratio of reads to writes should be recorded. A ratio of 1:1
2040 - the reads and writes MUST be of cache-line size and be cache-line aligned.
2041 - in NUMA architectures memory access SHOULD be local to the core's node.
2042 Whether only local memory or a mix of local and remote memory is used
2044 - the memory bandwidth (reads plus writes) used per-core MUST be recorded;
2045 the test MUST be run with a per-core memory bandwidth equal to half the
2046 maximum system memory bandwidth divided by the number of cores. The test
2047 MAY be run with other values for the per-core memory bandwidth.
2048 - the test MAY also be run with the memory intensive application running
2051 Under these conditions the DUT's 0% packet loss throughput is determined
2052 as per LTD.Throughput.RFC2544.PacketLossRatio.
2054 **Expected Result**:
2056 **Metrics Collected**:
2058 The following are the metrics collected for this test:
2060 - The DUT's 0% packet loss throughput in the presence of cache sharing and
2061 memory bandwidth between processes.
2066 -----------------------
2067 The general aim of these tests is to understand the capacity of the
2068 and speed with which the vswitch can accommodate new flows.
2072 Test ID: LTD.Activation.RFC2889.AddressCachingCapacity
2073 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2074 **Title**: RFC2889 Address Caching Capacity Test
2076 **Prerequisite Test**: N/A
2082 Please note this test is only applicable to virtual switches that are capable of
2083 MAC learning. The aim of this test is to determine the address caching
2084 capacity of the DUT for a constant load (fixed length frames at a fixed
2085 interval time). The selected frame sizes are those previously defined
2086 under `Default Test Parameters <#DefaultParams>`__.
2088 In order to run this test the aging time, that is the maximum time the
2089 DUT will keep a learned address in its flow table, and a set of initial
2090 addresses, whose value should be >= 1 and <= the max number supported by
2091 the implementation must be known. Please note that if the aging time is
2092 configurable it must be longer than the time necessary to produce frames
2093 from the external source at the specified rate. If the aging time is
2094 fixed the frame rate must be brought down to a value that the external
2095 source can produce in a time that is less than the aging time.
2097 Learning Frames should be sent from an external source to the DUT to
2098 install a number of flows. The Learning Frames must have a fixed
2099 destination address and must vary the source address of the frames. The
2100 DUT should install flows in its flow table based on the varying source
2101 addresses. Frames should then be transmitted from an external source at
2102 a suitable frame rate to see if the DUT has properly learned all of the
2103 addresses. If there is no frame loss and no flooding, the number of
2104 addresses sent to the DUT should be increased and the test is repeated
2105 until the max number of cached addresses supported by the DUT
2108 **Expected Result**:
2110 **Metrics collected**:
2112 The following are the metrics collected for this test:
2114 - Number of cached addresses supported by the DUT.
2115 - CPU and memory utilization may also be collected as part of this
2116 test, to determine the vSwitch's performance footprint on the system.
2118 **Deployment scenario**:
2120 - Physical → virtual switch → 2 x physical (one receiving, one listening).
2124 Test ID: LTD.Activation.RFC2889.AddressLearningRate
2125 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2126 **Title**: RFC2889 Address Learning Rate Test
2128 **Prerequisite Test**: LTD.Memory.RFC2889.AddressCachingCapacity
2134 Please note this test is only applicable to virtual switches that are capable of
2135 MAC learning. The aim of this test is to determine the rate of address
2136 learning of the DUT for a constant load (fixed length frames at a fixed
2137 interval time). The selected frame sizes are those previously defined
2138 under `Default Test Parameters <#DefaultParams>`__, traffic should be
2139 sent with each IPv4/IPv6 address incremented by one. The rate at which
2140 the DUT learns a new address should be measured. The maximum caching
2141 capacity from LTD.Memory.RFC2889.AddressCachingCapacity should be taken
2142 into consideration as the maximum number of addresses for which the
2143 learning rate can be obtained.
2145 **Expected Result**: It may be worthwhile to report the behaviour when
2146 operating beyond address capacity - some DUTs may be more friendly to
2147 new addresses than others.
2149 **Metrics collected**:
2151 The following are the metrics collected for this test:
2153 - The address learning rate of the DUT.
2155 **Deployment scenario**:
2157 - Physical → virtual switch → 2 x physical (one receiving, one listening).
2161 Coupling between control path and datapath Tests
2162 -------------------------------------------------------
2163 The following tests aim to determine how tightly coupled the datapath
2164 and the control path are within a virtual switch. The following list
2165 is not exhaustive but should indicate the type of tests that should be
2166 required. It is expected that more will be added.
2170 Test ID: LTD.CPDPCouplingFlowAddition
2171 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2172 **Title**: Control Path and Datapath Coupling
2174 **Prerequisite Test**:
2180 The aim of this test is to understand how exercising the DUT's control
2181 path affects datapath performance.
2183 Initially a certain number of flow table entries are installed in the
2184 vSwitch. Then over the duration of an RFC2544 throughput test
2185 flow-entries are added and removed at the rates specified below. No
2186 traffic is 'hitting' these flow-entries, they are simply added and
2189 The test MUST be repeated with the following initial number of
2190 flow-entries installed: - < 10 - 1000 - 100,000 - 10,000,000 (or the
2191 maximum supported number of flow-entries)
2193 The test MUST be repeated with the following rates of flow-entry
2194 addition and deletion per second: - 0 - 1 (i.e. 1 addition plus 1
2195 deletion) - 100 - 10,000
2197 **Expected Result**:
2199 **Metrics Collected**:
2201 The following are the metrics collected for this test:
2203 - The maximum forwarding rate in Frames Per Second (FPS) and Mbps of
2205 - The average latency of the traffic flow when passing through the DUT
2206 (if testing for latency, note that this average is different from the
2207 test specified in Section 26.3 of
2208 `RFC2544 <https://www.rfc-editor.org/rfc/rfc2544.txt>`__).
2209 - CPU and memory utilization may also be collected as part of this
2210 test, to determine the vSwitch's performance footprint on the system.
2212 **Deployment scenario**:
2214 - Physical → virtual switch → physical.
2218 CPU and memory consumption
2219 ---------------------------------
2220 The following tests will profile a virtual switch's CPU and memory
2221 utilization under various loads and circumstances. The following
2222 list is not exhaustive but should indicate the type of tests that
2223 should be required. It is expected that more will be added.
2227 Test ID: LTD.CPU.RFC2544.0PacketLoss
2228 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2229 **Title**: RFC 2544 0% Loss Compute Test
2231 **Prerequisite Test**:
2237 The aim of this test is to understand the overall performance of the
2238 system when a CPU intensive application is run on the same DUT as the
2239 Virtual Switch. For each frame size, an
2240 LTD.Throughput.RFC2544.PacketLossRatio (0% Packet Loss) test should be
2241 performed. Throughout the entire test a CPU intensive application should
2242 be run on all cores on the system not in use by the Virtual Switch. For
2243 NUMA system only cores on the same NUMA node are loaded.
2245 It is recommended that stress-ng be used for loading the non-Virtual
2246 Switch cores but any stress tool MAY be used.
2248 **Expected Result**:
2250 **Metrics Collected**:
2252 The following are the metrics collected for this test:
2254 - CPU utilization of the cores running the Virtual Switch.
2255 - The number of identity of the cores allocated to the Virtual Switch.
2256 - The configuration of the stress tool (for example the command line
2257 parameters used to start it.)
2261 Summary List of Tests
2262 ----------------------------
2265 - Test ID: LTD.Throughput.RFC2544.PacketLossRatio
2266 - Test ID: LTD.Throughput.RFC2544.PacketLossRatioFrameModification
2267 - Test ID: LTD.Throughput.RFC2544.Profile
2268 - Test ID: LTD.Throughput.RFC2544.SystemRecoveryTime
2269 - Test ID: LTD.Throughput.RFC2544.BackToBackFrames
2270 - Test ID: LTD.Throughput.RFC2889.Soak
2271 - Test ID: LTD.Throughput.RFC2889.SoakFrameModification
2272 - Test ID: LTD.Throughput.RFC6201.ResetTime
2273 - Test ID: LTD.Throughput.RFC2889.MaxForwardingRate
2274 - Test ID: LTD.Throughput.RFC2889.ForwardPressure
2275 - Test ID: LTD.Throughput.RFC2889.ErrorFramesFiltering
2276 - Test ID: LTD.Throughput.RFC2889.BroadcastFrameForwarding
2278 2. Packet Latency tests
2280 - Test ID: LTD.PacketLatency.InitialPacketProcessingLatency
2281 - Test ID: LTD.PacketDelayVariation.RFC3393.Soak
2283 3. Scalability tests
2285 - Test ID: LTD.Scalability.RFC2544.0PacketLoss
2286 - Test ID: LTD.MemoryBandwidth.RFC2544.0PacketLoss.Scalability
2290 - Test ID: LTD.Activation.RFC2889.AddressCachingCapacity
2291 - Test ID: LTD.Activation.RFC2889.AddressLearningRate
2293 5. Coupling between control path and datapath Tests
2295 - Test ID: LTD.CPDPCouplingFlowAddition
2297 6. CPU and memory consumption
2299 - Test ID: LTD.CPU.RFC2544.0PacketLoss
Code Review / vswitchperf.git / blob