1 <?xml version="1.0" encoding="US-ASCII"?>
2 <!DOCTYPE rfc SYSTEM "rfc2629.dtd">
4 <?rfc tocompact="yes"?>
6 <?rfc tocindent="yes"?>
12 <?rfc subcompact="no"?>
13 <rfc category="info" docName="draft-ietf-bmwg-vswitch-opnfv-01"
16 <title abbrev="Benchmarking vSwitches">Benchmarking Virtual Switches in
19 <author fullname="Maryam Tahhan" initials="M." surname="Tahhan">
20 <organization>Intel</organization>
39 <email>maryam.tahhan@intel.com</email>
45 <author fullname="Billy O'Mahony" initials="B." surname="O'Mahony">
46 <organization>Intel</organization>
65 <email>billy.o.mahony@intel.com</email>
71 <author fullname="Al Morton" initials="A." surname="Morton">
72 <organization>AT&T Labs</organization>
76 <street>200 Laurel Avenue South</street>
78 <city>Middletown,</city>
84 <country>USA</country>
87 <phone>+1 732 420 1571</phone>
89 <facsimile>+1 732 368 1192</facsimile>
91 <email>acmorton@att.com</email>
93 <uri>http://home.comcast.net/~acmacm/</uri>
97 <date day="10" month="October" year="2016"/>
100 <t>This memo describes the progress of the Open Platform for NFV (OPNFV)
101 project on virtual switch performance "VSWITCHPERF". This project
102 intends to build on the current and completed work of the Benchmarking
103 Methodology Working Group in IETF, by referencing existing literature.
104 The Benchmarking Methodology Working Group has traditionally conducted
105 laboratory characterization of dedicated physical implementations of
106 internetworking functions. Therefore, this memo begins to describe the
107 additional considerations when virtual switches are implemented in
108 general-purpose hardware. The expanded tests and benchmarks are also
109 influenced by the OPNFV mission to support virtualization of the "telco"
113 <note title="Requirements Language">
114 <t>The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
115 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
116 document are to be interpreted as described in <xref
117 target="RFC2119">RFC 2119</xref>.</t>
124 <section title="Introduction">
125 <t>Benchmarking Methodology Working Group (BMWG) has traditionally
126 conducted laboratory characterization of dedicated physical
127 implementations of internetworking functions. The Black-box Benchmarks
128 of Throughput, Latency, Forwarding Rates and others have served our
129 industry for many years. Now, Network Function Virtualization (NFV) has
130 the goal to transform how internetwork functions are implemented, and
131 therefore has garnered much attention.</t>
133 <t>This memo summarizes the progress of the Open Platform for NFV
134 (OPNFV) project on virtual switch performance characterization,
135 "VSWITCHPERF", through the Brahmaputra (second) release <xref
136 target="BrahRel"/>. This project intends to build on the current and
137 completed work of the Benchmarking Methodology Working Group in IETF, by
138 referencing existing literature. For example, currently the most often
139 referenced RFC is <xref target="RFC2544"/> (which depends on <xref
140 target="RFC1242"/>) and foundation of the benchmarking work in OPNFV is
141 common and strong.</t>
144 https://wiki.opnfv.org/characterize_vswitch_performance_for_telco_nfv_use_cases
145 for more background, and the OPNFV website for general information:
146 https://www.opnfv.org/</t>
148 <t>The authors note that OPNFV distinguishes itself from other open
149 source compute and networking projects through its emphasis on existing
150 "telco" services as opposed to cloud-computing. There are many ways in
151 which telco requirements have different emphasis on performance
152 dimensions when compared to cloud computing: support for and transfer of
153 isochronous media streams is one example.</t>
155 <t>Note also that the move to NFV Infrastructure has resulted in many
156 new benchmarking initiatives across the industry. The authors are
157 currently doing their best to maintain alignment with many other
158 projects, and this Internet Draft is one part of the efforts. We
159 acknowledge the early work in <xref
160 target="I-D.huang-bmwg-virtual-network-performance"/>, and useful
161 discussion with the authors.</t>
164 <section title="Scope">
165 <t>The primary purpose and scope of the memo is to inform the industry
166 of work-in-progress that builds on the body of extensive BMWG literature
167 and experience, and describe the extensions needed for benchmarking
168 virtual switches. Inital feedback indicates that many of these
169 extensions may be applicable beyond the current scope (to hardware
170 switches in the NFV Infrastructure and to virtual routers, for example).
171 Additionally, this memo serves as a vehicle to include more detail and
172 commentary from BMWG and other Open Source communities, under BMWG's
173 chartered work to characterize the NFV Infrastructure (a virtual switch
174 is an important aspect of that infrastructure).</t>
176 <t>The benchmarking covered in this memo should be applicable to many
177 types of vswitches, and remain vswitch-agnostic to great degree. There
178 has been no attempt to track and test all features of any specific
179 vswitch implementation.</t>
182 <section title="Benchmarking Considerations">
183 <t>This section highlights some specific considerations (from <xref
184 target="I-D.ietf-bmwg-virtual-net"/>)related to Benchmarks for virtual
185 switches. The OPNFV project is sharing its present view on these areas,
186 as they develop their specifications in the Level Test Design (LTD)
189 <section title="Comparison with Physical Network Functions">
190 <t>To compare the performance of virtual designs and implementations
191 with their physical counterparts, identical benchmarks are needed.
192 BMWG has developed specifications for many network functions this memo
193 re-uses existing benchmarks through references, and expands them
194 during development of new methods. A key configuration aspect is the
195 number of parallel cores required to achieve comparable performance
196 with a given physical device, or whether some limit of scale was
197 reached before the cores could achieve the comparable level.</t>
199 <t>It's unlikely that the virtual switch will be the only application
200 running on the SUT, so CPU utilization, Cache utilization, and Memory
201 footprint should also be recorded for the virtual implementations of
202 internetworking functions.</t>
205 <section title="Continued Emphasis on Black-Box Benchmarks">
206 <t>External observations remain essential as the basis for Benchmarks.
207 Internal observations with fixed specification and interpretation will
208 be provided in parallel to assist the development of operations
209 procedures when the technology is deployed.</t>
212 <section title="New Configuration Parameters">
213 <t>A key consideration when conducting any sort of benchmark is trying
214 to ensure the consistency and repeatability of test results. When
215 benchmarking the performance of a vSwitch there are many factors that
216 can affect the consistency of results, one key factor is matching the
217 various hardware and software details of the SUT. This section lists
218 some of the many new parameters which this project believes are
219 critical to report in order to achieve repeatability.</t>
221 <t>Hardware details including:</t>
223 <t><list style="symbols">
224 <t>Platform details</t>
226 <t>Processor details</t>
228 <t>Memory information (type and size)</t>
230 <t>Number of enabled cores</t>
232 <t>Number of cores used for the test</t>
234 <t>Number of physical NICs, as well as their details
235 (manufacturer, versions, type and the PCI slot they are plugged
238 <t>NIC interrupt configuration</t>
240 <t>BIOS version, release date and any configurations that were
243 <t>CPU microcode level</t>
245 <t>Memory DIMM configurations (quad rank performance may not be
246 the same as dual rank) in size, freq and slot locations</t>
248 <t>PCI configuration parameters (payload size, early ack
251 <t>Power management at all levels (ACPI sleep states, processor
253 </list>Software details including:</t>
255 <t><list style="symbols">
256 <t>OS parameters and behavior (text vs graphical no one typing at
257 the console on one system)</t>
259 <t>OS version (for host and VNF)</t>
261 <t>Kernel version (for host and VNF)</t>
263 <t>GRUB boot parameters (for host and VNF)</t>
265 <t>Hypervisor details (Type and version)</t>
267 <t>Selected vSwitch, version number or commit id used</t>
269 <t>vSwitch launch command line if it has been parameterised</t>
271 <t>Memory allocation to the vSwitch</t>
273 <t>which NUMA node it is using, and how many memory channels</t>
275 <t>DPDK or any other SW dependency version number or commit id
278 <t>Memory allocation to a VM - if it's from Hugpages/elsewhere</t>
280 <t>VM storage type: snapshot/independent persistent/independent
285 <t>Number of Virtual NICs (vNICs), versions, type and driver</t>
287 <t>Number of virtual CPUs and their core affinity on the host</t>
289 <t>Number vNIC interrupt configuration</t>
291 <t>Thread affinitization for the applications (including the
292 vSwitch itself) on the host</t>
294 <t>Details of Resource isolation, such as CPUs designated for
295 Host/Kernel (isolcpu) and CPUs designated for specific processes
296 (taskset). - Test duration. - Number of flows.</t>
299 <t>Test Traffic Information:<list style="symbols">
300 <t>Traffic type - UDP, TCP, IMIX / Other</t>
304 <t>Deployment Scenario</t>
310 <section title="Flow classification">
311 <t>Virtual switches group packets into flows by processing and
312 matching particular packet or frame header information, or by matching
313 packets based on the input ports. Thus a flow can be thought of a
314 sequence of packets that have the same set of header field values
315 (5-tuple) or have arrived on the same port. Performance results can
316 vary based on the parameters the vSwitch uses to match for a flow. The
317 recommended flow classification parameters for any vSwitch performance
318 tests are: the input port, the source IP address, the destination IP
319 address and the Ethernet protocol type field. It is essential to
320 increase the flow timeout time on a vSwitch before conducting any
321 performance tests that do not measure the flow setup time. Normally
322 the first packet of a particular stream will install the flow in the
323 virtual switch which adds an additional latency, subsequent packets of
324 the same flow are not subject to this latency if the flow is already
325 installed on the vSwitch.</t>
328 <section title="Benchmarks using Baselines with Resource Isolation">
329 <t>This outline describes measurement of baseline with isolated
330 resources at a high level, which is the intended approach at this
333 <t><list style="numbers">
334 <t>Baselines: <list style="symbols">
335 <t>Optional: Benchmark platform forwarding capability without
336 a vswitch or VNF for at least 72 hours (serves as a means of
337 platform validation and a means to obtain the base performance
338 for the platform in terms of its maximum forwarding rate and
340 <preamble>Benchmark platform forwarding
341 capability</preamble>
343 <artwork align="right"><![CDATA[ __
344 +--------------------------------------------------+ |
345 | +------------------------------------------+ | |
347 | | Simple Forwarding App | | Host
349 | +------------------------------------------+ | |
351 +---+------------------------------------------+---+ __|
355 +--------------------------------------------------+
357 | traffic generator |
359 +--------------------------------------------------+]]></artwork>
364 <t>Benchmark VNF forwarding capability with direct
365 connectivity (vSwitch bypass, e.g., SR/IOV) for at least 72
366 hours (serves as a means of VNF validation and a means to
367 obtain the base performance for the VNF in terms of its
368 maximum forwarding rate and latency). The metrics gathered
369 from this test will serve as a key comparison point for
370 vSwitch bypass technologies performance and vSwitch
371 performance. <figure align="right">
372 <preamble>Benchmark VNF forwarding capability</preamble>
374 <artwork><![CDATA[ __
375 +--------------------------------------------------+ |
376 | +------------------------------------------+ | |
380 | +------------------------------------------+ | |
381 | | Passthrough/SR-IOV | | Host
382 | +------------------------------------------+ | |
384 +---+------------------------------------------+---+ __|
388 +--------------------------------------------------+
390 | traffic generator |
392 +--------------------------------------------------+]]></artwork>
397 <t>Benchmarking with isolated resources alone, with other
398 resources (both HW&SW) disabled Example, vSw and VM are
401 <t>Benchmarking with isolated resources alone, leaving some
404 <t>Benchmark with isolated resources and all resources
408 <t>Next Steps<list style="symbols">
409 <t>Limited sharing</t>
411 <t>Production scenarios</t>
413 <t>Stressful scenarios</t>
419 <section title="VSWITCHPERF Specification Summary">
420 <t>The overall specification in preparation is referred to as a Level
421 Test Design (LTD) document, which will contain a suite of performance
422 tests. The base performance tests in the LTD are based on the
423 pre-existing specifications developed by BMWG to test the performance of
424 physical switches. These specifications include:</t>
426 <t><list style="symbols">
427 <t><xref target="RFC2544"/> Benchmarking Methodology for Network
428 Interconnect Devices</t>
430 <t><xref target="RFC2889"/> Benchmarking Methodology for LAN
433 <t><xref target="RFC6201"/> Device Reset Characterization</t>
435 <t><xref target="RFC5481"/> Packet Delay Variation Applicability
439 <t>Some of the above/newer RFCs are being applied in benchmarking for
440 the first time, and represent a development challenge for test equipment
441 developers. Fortunately, many members of the testing system community
442 have engaged on the VSPERF project, including an open source test
445 <t>In addition to this, the LTD also re-uses the terminology defined
448 <t><list style="symbols">
449 <t><xref target="RFC2285"/> Benchmarking Terminology for LAN
450 Switching Devices</t>
452 <t><xref target="RFC5481"/> Packet Delay Variation Applicability
458 <t>Specifications to be included in future updates of the LTD
459 include:<list style="symbols">
460 <t><xref target="RFC3918"/> Methodology for IP Multicast
463 <t><xref target="RFC4737"/> Packet Reordering Metrics</t>
466 <t>As one might expect, the most fundamental internetworking
467 characteristics of Throughput and Latency remain important when the
468 switch is virtualized, and these benchmarks figure prominently in the
471 <t>When considering characteristics important to "telco" network
472 functions, we must begin to consider additional performance metrics. In
473 this case, the project specifications have referenced metrics from the
474 IETF IP Performance Metrics (IPPM) literature. This means that the <xref
475 target="RFC2544"/> test of Latency is replaced by measurement of a
476 metric derived from IPPM's <xref target="RFC2679"/>, where a set of
477 statistical summaries will be provided (mean, max, min, etc.). Further
478 metrics planned to be benchmarked include packet delay variation as
479 defined by <xref target="RFC5481"/> , reordering, burst behaviour, DUT
480 availability, DUT capacity and packet loss in long term testing at
481 Throughput level, where some low-level of background loss may be present
482 and characterized.</t>
484 <t>Tests have been (or will be) designed to collect the metrics
487 <t><list style="symbols">
488 <t>Throughput Tests to measure the maximum forwarding rate (in
489 frames per second or fps) and bit rate (in Mbps) for a constant load
490 (as defined by <xref target="RFC1242"/>) without traffic loss.</t>
492 <t>Packet and Frame Delay Distribution Tests to measure average, min
493 and max packet and frame delay for constant loads.</t>
495 <t>Packet Delay Tests to understand latency distribution for
496 different packet sizes and over an extended test run to uncover
499 <t>Scalability Tests to understand how the virtual switch performs
500 as the number of flows, active ports, complexity of the forwarding
501 logic’s configuration… it has to deal with
504 <t>Stream Performance Tests (TCP, UDP) to measure bulk data transfer
505 performance, i.e. how fast systems can send and receive data through
508 <t>Control Path and Datapath Coupling Tests, to understand how
509 closely coupled the datapath and the control path are as well as the
510 effect of this coupling on the performance of the DUT (example:
511 delay of the initial packet of a flow).</t>
513 <t>CPU and Memory Consumption Tests to understand the virtual
514 switch’s footprint on the system, usually conducted as
515 auxiliary measurements with benchmarks above. They include: CPU
516 utilization, Cache utilization and Memory footprint.</t>
518 <t>The so-called "Soak" tests, where the selected test is conducted
519 over a long period of time (with an ideal duration of 24 hours, but
520 only long enough to determine that stability issues exist when
521 found; there is no requirement to continue a test when a DUT
522 exhibits instability over time). The key performance characteristics
523 and benchmarks for a DUT are determined (using short duration tests)
524 prior to conducting soak tests. The purpose of soak tests is to
525 capture transient changes in performance which may occur due to
526 infrequent processes, memory leaks, or the low probability
527 coincidence of two or more processes. The stability of the DUT is
528 the paramount consideration, so performance must be evaluated
529 periodically during continuous testing, and this results in use of
530 <xref target="RFC2889"/> Frame Rate metrics instead of <xref
531 target="RFC2544"/> Throughput (which requires stopping traffic to
532 allow time for all traffic to exit internal queues), for
536 <t>Future/planned test specs include:<list style="symbols">
537 <t>Request/Response Performance Tests (TCP, UDP) which measure the
538 transaction rate through the switch.</t>
540 <t>Noisy Neighbour Tests, to understand the effects of resource
541 sharing on the performance of a virtual switch.</t>
543 <t>Tests derived from examination of ETSI NFV Draft GS IFA003
544 requirements <xref target="IFA003"/> on characterization of
545 acceleration technologies applied to vswitches.</t>
546 </list>The flexibility of deployment of a virtual switch within a
547 network means that the BMWG IETF existing literature needs to be used to
548 characterize the performance of a switch in various deployment
549 scenarios. The deployment scenarios under consideration include:</t>
552 <preamble>Physical port to virtual switch to physical
555 <artwork><![CDATA[ __
556 +--------------------------------------------------+ |
557 | +--------------------+ | |
560 | +--------------+ +--------------+ | |
561 | | phy port | vSwitch | phy port | | |
562 +---+--------------+------------+--------------+---+ __|
566 +--------------------------------------------------+
568 | traffic generator |
570 +--------------------------------------------------+]]></artwork>
574 <preamble>Physical port to virtual switch to VNF to virtual switch
575 to physical port</preamble>
577 <artwork><![CDATA[ __
578 +---------------------------------------------------+ |
580 | +-------------------------------------------+ | |
581 | | Application | | |
582 | +-------------------------------------------+ | |
586 | +---------------+ +---------------+ | |
587 | | logical port 0| | logical port 1| | |
588 +---+---------------+-----------+---------------+---+ __|
592 +---+---------------+----------+---------------+---+ |
593 | | logical port 0| | logical port 1| | |
594 | +---------------+ +---------------+ | |
598 | +--------------+ +--------------+ | |
599 | | phy port | vSwitch | phy port | | |
600 +---+--------------+------------+--------------+---+ __|
604 +--------------------------------------------------+
606 | traffic generator |
608 +--------------------------------------------------+]]></artwork>
610 <preamble>Physical port to virtual switch to VNF to virtual switch
611 to VNF to virtual switch to physical port</preamble>
613 <artwork><![CDATA[ __
614 +----------------------+ +----------------------+ |
615 | Guest 1 | | Guest 2 | |
616 | +---------------+ | | +---------------+ | |
617 | | Application | | | | Application | | |
618 | +---------------+ | | +---------------+ | |
620 | | v | | | v | | Guests
621 | +---------------+ | | +---------------+ | |
622 | | logical ports | | | | logical ports | | |
623 | | 0 1 | | | | 0 1 | | |
624 +---+---------------+--+ +---+---------------+--+__|
628 +---+---------------+---------+---------------+--+ |
629 | | 0 1 | | 3 4 | | |
630 | | logical ports | | logical ports | | |
631 | +---------------+ +---------------+ | |
633 | | |-----------------| v | |
634 | +--------------+ +--------------+ | |
635 | | phy ports | vSwitch | phy ports | | |
636 +---+--------------+----------+--------------+---+_|
640 +--------------------------------------------------+
642 | traffic generator |
644 +--------------------------------------------------+]]></artwork>
646 <preamble>Physical port to virtual switch to VNF</preamble>
648 <artwork><![CDATA[ __
649 +---------------------------------------------------+ |
651 | +-------------------------------------------+ | |
652 | | Application | | |
653 | +-------------------------------------------+ | |
657 | +---------------+ | |
658 | | logical port 0| | |
659 +---+---------------+-------------------------------+ __|
663 +---+---------------+------------------------------+ |
664 | | logical port 0| | |
665 | +---------------+ | |
669 | +--------------+ | |
670 | | phy port | vSwitch | |
671 +---+--------------+------------ -------------- ---+ __|
675 +--------------------------------------------------+
677 | traffic generator |
679 +--------------------------------------------------+]]></artwork>
681 <preamble>VNF to virtual switch to physical port</preamble>
683 <artwork><![CDATA[ __
684 +---------------------------------------------------+ |
686 | +-------------------------------------------+ | |
687 | | Application | | |
688 | +-------------------------------------------+ | |
692 | +---------------+ | |
693 | | logical port | | |
694 +-------------------------------+---------------+---+ __|
698 +------------------------------+---------------+---+ |
699 | | logical port | | |
700 | +---------------+ | |
704 | +--------------+ | |
705 | vSwitch | phy port | | |
706 +-------------------------------+--------------+---+ __|
710 +--------------------------------------------------+
712 | traffic generator |
714 +--------------------------------------------------+]]></artwork>
716 <preamble>VNF to virtual switch to VNF</preamble>
718 <artwork><![CDATA[ __
719 +----------------------+ +----------------------+ |
720 | Guest 1 | | Guest 2 | |
721 | +---------------+ | | +---------------+ | |
722 | | Application | | | | Application | | |
723 | +---------------+ | | +---------------+ | |
726 | +---------------+ | | +---------------+ | |
727 | | logical ports | | | | logical ports | | |
728 | | 0 | | | | 0 | | |
729 +---+---------------+--+ +---+---------------+--+__|
733 +---+---------------+---------+---------------+--+ |
735 | | logical ports | | logical ports | | |
736 | +---------------+ +---------------+ | |
738 | L-----------------+ | |
741 +------------------------------------------------+_|]]></artwork>
744 <t>A set of Deployment Scenario figures is available on the VSPERF Test
745 Methodology Wiki page <xref target="TestTopo"/>.</t>
748 <section title="3x3 Matrix Coverage">
749 <t>This section organizes the many existing test specifications into the
750 "3x3" matrix (introduced in <xref target="I-D.ietf-bmwg-virtual-net"/>).
751 Because the LTD specification ID names are quite long, this section is
752 organized into lists for each occupied cell of the matrix (not all are
753 occupied, also the matrix has grown to 3x4 to accommodate scale metrics
754 when displaying the coverage of many metrics/benchmarks). The current
755 version of the LTD specification is available <xref target="LTD"/>.</t>
757 <t>The tests listed below assess the activation of paths in the data
758 plane, rather than the control plane.</t>
760 <t>A complete list of tests with short summaries is available on the
761 VSPERF "LTD Test Spec Overview" Wiki page <xref target="LTDoverV"/>.</t>
763 <section title="Speed of Activation">
764 <t><list style="symbols">
765 <t>Activation.RFC2889.AddressLearningRate</t>
767 <t>PacketLatency.InitialPacketProcessingLatency</t>
771 <section title="Accuracy of Activation section">
772 <t><list style="symbols">
773 <t>CPDP.Coupling.Flow.Addition</t>
777 <section title="Reliability of Activation">
778 <t><list style="symbols">
779 <t>Throughput.RFC2544.SystemRecoveryTime</t>
781 <t>Throughput.RFC2544.ResetTime</t>
785 <section title="Scale of Activation">
786 <t><list style="symbols">
787 <t>Activation.RFC2889.AddressCachingCapacity</t>
791 <section title="Speed of Operation">
792 <t><list style="symbols">
793 <t>Throughput.RFC2544.PacketLossRate</t>
795 <t>CPU.RFC2544.0PacketLoss</t>
797 <t>Throughput.RFC2544.PacketLossRateFrameModification</t>
799 <t>Throughput.RFC2544.BackToBackFrames</t>
801 <t>Throughput.RFC2889.MaxForwardingRate</t>
803 <t>Throughput.RFC2889.ForwardPressure</t>
805 <t>Throughput.RFC2889.BroadcastFrameForwarding</t>
809 <section title="Accuracy of Operation">
810 <t><list style="symbols">
811 <t>Throughput.RFC2889.ErrorFramesFiltering</t>
813 <t>Throughput.RFC2544.Profile</t>
817 <section title="Reliability of Operation">
818 <t><list style="symbols">
819 <t>Throughput.RFC2889.Soak</t>
821 <t>Throughput.RFC2889.SoakFrameModification</t>
823 <t>PacketDelayVariation.RFC3393.Soak</t>
827 <section title="Scalability of Operation">
828 <t><list style="symbols">
829 <t>Scalability.RFC2544.0PacketLoss</t>
831 <t>MemoryBandwidth.RFC2544.0PacketLoss.Scalability</t>
835 <section title="Summary">
837 <artwork><![CDATA[|------------------------------------------------------------------------|
839 | | SPEED | ACCURACY | RELIABILITY | SCALE |
841 |------------------------------------------------------------------------|
843 | Activation | X | X | X | X |
845 |------------------------------------------------------------------------|
847 | Operation | X | X | X | X |
849 |------------------------------------------------------------------------|
851 | De-activation | | | | |
853 |------------------------------------------------------------------------|]]></artwork>
858 <section title="Security Considerations">
859 <t>Benchmarking activities as described in this memo are limited to
860 technology characterization of a Device Under Test/System Under Test
861 (DUT/SUT) using controlled stimuli in a laboratory environment, with
862 dedicated address space and the constraints specified in the sections
865 <t>The benchmarking network topology will be an independent test setup
866 and MUST NOT be connected to devices that may forward the test traffic
867 into a production network, or misroute traffic to the test management
870 <t>Further, benchmarking is performed on a "black-box" basis, relying
871 solely on measurements observable external to the DUT/SUT.</t>
873 <t>Special capabilities SHOULD NOT exist in the DUT/SUT specifically for
874 benchmarking purposes. Any implications for network security arising
875 from the DUT/SUT SHOULD be identical in the lab and in production
879 <section anchor="IANA" title="IANA Considerations">
880 <t>No IANA Action is requested at this time.</t>
883 <section title="Acknowledgements">
884 <t>The authors appreciate and acknowledge comments from Scott Bradner,
885 Marius Georgescu, Ramki Krishnan, Doug Montgomery, Martin Klozik,
886 Christian Trautman, and others for their reviews.</t>
891 <references title="Normative References">
894 <?rfc include="reference.RFC.2119"?>
898 <?rfc include="reference.RFC.2330"?>
900 <?rfc include='reference.RFC.2544'?>
902 <?rfc include="reference.RFC.2679"?>
904 <?rfc include='reference.RFC.2680'?>
906 <?rfc include='reference.RFC.3393'?>
908 <?rfc include='reference.RFC.3432'?>
910 <?rfc include='reference.RFC.2681'?>
912 <?rfc include='reference.RFC.5905'?>
914 <?rfc include='reference.RFC.4689'?>
916 <?rfc include='reference.RFC.4737'?>
918 <?rfc include='reference.RFC.5357'?>
920 <?rfc include='reference.RFC.2889'?>
922 <?rfc include='reference.RFC.3918'?>
924 <?rfc include='reference.RFC.6201'?>
926 <?rfc include='reference.RFC.2285'?>
928 <reference anchor="NFV.PER001">
930 <title>Network Function Virtualization: Performance and Portability
931 Best Practices</title>
933 <author fullname="ETSI NFV" initials="" surname="">
937 <date month="June" year="2014"/>
940 <seriesInfo name="Group Specification"
941 value="ETSI GS NFV-PER 001 V1.1.1 (2014-06)"/>
947 <references title="Informative References">
948 <?rfc include='reference.RFC.1242'?>
950 <?rfc include='reference.RFC.5481'?>
952 <?rfc include='reference.RFC.6049'?>
954 <?rfc include='reference.RFC.6248'?>
956 <?rfc include='reference.RFC.6390'?>
958 <?rfc include='reference.I-D.ietf-bmwg-virtual-net'?>
960 <?rfc include='reference.I-D.huang-bmwg-virtual-network-performance'?>
962 <reference anchor="TestTopo">
964 <title>Test Topologies
965 https://wiki.opnfv.org/vsperf/test_methodology</title>
975 <reference anchor="LTDoverV">
977 <title>LTD Test Spec Overview
978 https://wiki.opnfv.org/wiki/vswitchperf_test_spec_review</title>
988 <reference anchor="LTD">
990 <title>LTD Test Specification
991 http://artifacts.opnfv.org/vswitchperf/brahmaputra/docs/requirements/index.html</title>
1001 <reference anchor="BrahRel">
1003 <title>Brahmaputra, Second OPNFV Release
1004 https://www.opnfv.org/brahmaputra</title>
1014 <reference anchor="IFA003">
1016 <title>https://docbox.etsi.org/ISG/NFV/Open/Drafts/IFA003_Acceleration_-_vSwitch_Spec/</title>