1 .. This work is licensed under a Creative Commons Attribution 4.0 International License.
2 .. http://creativecommons.org/licenses/by/4.0
3 .. (c) OPNFV, Intel Corporation, AT&T and others.
5 vSwitchPerf test suites userguide
6 ---------------------------------
11 VSPERF requires a traffic generators to run tests, automated traffic gen
12 support in VSPERF includes:
14 - IXIA traffic generator (IxNetwork hardware) and a machine that runs the IXIA
16 - Spirent traffic generator (TestCenter hardware chassis or TestCenter virtual
17 in a VM) and a VM to run the Spirent Virtual Deployment Service image,
18 formerly known as "Spirent LabServer".
19 - Xena Network traffic generator (Xena hardware chassis) that houses the Xena
20 Traffic generator modules.
21 - Moongen software traffic generator. Requires a separate machine running
22 moongen to execute packet generation.
24 If you want to use another traffic generator, please select the :ref:`trafficgen-dummy`
30 To see the supported Operating Systems, vSwitches and system requirements,
31 please follow the `installation instructions <vsperf-installation>`.
33 Traffic Generator Setup
34 ^^^^^^^^^^^^^^^^^^^^^^^
36 Follow the `Traffic generator instructions <trafficgen-installation>` to
37 install and configure a suitable traffic generator.
39 Cloning and building src dependencies
40 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
42 In order to run VSPERF, you will need to download DPDK and OVS. You can
43 do this manually and build them in a preferred location, OR you could
44 use vswitchperf/src. The vswitchperf/src directory contains makefiles
45 that will allow you to clone and build the libraries that VSPERF depends
46 on, such as DPDK and OVS. To clone and build simply:
48 .. code-block:: console
53 VSPERF can be used with stock OVS (without DPDK support). When build
54 is finished, the libraries are stored in src_vanilla directory.
56 The 'make' builds all options in src:
59 * OVS with vhost_user as the guest access method (with DPDK support)
61 The vhost_user build will reside in src/ovs/
62 The Vanilla OVS build will reside in vswitchperf/src_vanilla
64 To delete a src subdirectory and its contents to allow you to re-clone simply
67 .. code-block:: console
71 Configure the ``./conf/10_custom.conf`` file
72 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
74 The ``10_custom.conf`` file is the configuration file that overrides
75 default configurations in all the other configuration files in ``./conf``
76 The supplied ``10_custom.conf`` file **MUST** be modified, as it contains
77 configuration items for which there are no reasonable default values.
79 The configuration items that can be added is not limited to the initial
80 contents. Any configuration item mentioned in any .conf file in
81 ``./conf`` directory can be added and that item will be overridden by
82 the custom configuration value.
84 Further details about configuration files evaluation and special behaviour
85 of options with ``GUEST_`` prefix could be found at :ref:`design document
86 <design-configuration>`.
88 Using a custom settings file
89 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
91 If your ``10_custom.conf`` doesn't reside in the ``./conf`` directory
92 of if you want to use an alternative configuration file, the file can
93 be passed to ``vsperf`` via the ``--conf-file`` argument.
95 .. code-block:: console
97 $ ./vsperf --conf-file <path_to_custom_conf> ...
99 Note that configuration passed in via the environment (``--load-env``)
100 or via another command line argument will override both the default and
101 your custom configuration files. This "priority hierarchy" can be
102 described like so (1 = max priority):
104 1. Testcase definition section ``Parameters``
105 2. Command line arguments
106 3. Environment variables
107 4. Configuration file(s)
109 Further details about configuration files evaluation and special behaviour
110 of options with ``GUEST_`` prefix could be found at :ref:`design document
111 <design-configuration>`.
113 .. _overriding-parameters-documentation:
115 Overriding values defined in configuration files
116 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
118 The configuration items can be overridden by command line argument
119 ``--test-params``. In this case, the configuration items and
120 their values should be passed in form of ``item=value`` and separated
127 $ ./vsperf --test-params "TRAFFICGEN_DURATION=10;TRAFFICGEN_PKT_SIZES=(128,);" \
128 "GUEST_LOOPBACK=['testpmd','l2fwd']" pvvp_tput
130 The second option is to override configuration items by ``Parameters`` section
131 of the test case definition. The configuration items can be added into ``Parameters``
132 dictionary with their new values. These values will override values defined in
133 configuration files or specified by ``--test-params`` command line argument.
139 "Parameters" : {'TRAFFICGEN_PKT_SIZES' : (128,),
140 'TRAFFICGEN_DURATION' : 10,
141 'GUEST_LOOPBACK' : ['testpmd','l2fwd'],
144 **NOTE:** In both cases, configuration item names and their values must be specified
145 in the same form as they are defined inside configuration files. Parameter names
146 must be specified in uppercase and data types of original and new value must match.
147 Python syntax rules related to data types and structures must be followed.
148 For example, parameter ``TRAFFICGEN_PKT_SIZES`` above is defined as a tuple
149 with a single value ``128``. In this case trailing comma is mandatory, otherwise
150 value can be wrongly interpreted as a number instead of a tuple and vsperf
151 execution would fail. Please check configuration files for default values and their
152 types and use them as a basis for any customized values. In case of any doubt, please
153 check official python documentation related to data structures like tuples, lists
156 **NOTE:** Vsperf execution will terminate with runtime error in case, that unknown
157 parameter name is passed via ``--test-params`` CLI argument or defined in ``Parameters``
158 section of test case definition. It is also forbidden to redefine a value of
159 ``TEST_PARAMS`` configuration item via CLI or ``Parameters`` section.
164 VSPERF uses a VM image called vloop_vnf for looping traffic in the deployment
165 scenarios involving VMs. The image can be downloaded from
166 `<http://artifacts.opnfv.org/>`__.
168 Please see the installation instructions for information on :ref:`vloop-vnf`
176 A Kernel Module that provides OSI Layer 2 Ipv4 termination or forwarding with
177 support for Destination Network Address Translation (DNAT) for both the MAC and
178 IP addresses. l2fwd can be found in <vswitchperf_dir>/src/l2fwd
183 All examples inside these docs assume, that user is inside the VSPERF
184 directory. VSPERF can be executed from any directory.
186 Before running any tests make sure you have root permissions by adding
187 the following line to /etc/sudoers:
189 .. code-block:: console
191 username ALL=(ALL) NOPASSWD: ALL
193 username in the example above should be replaced with a real username.
195 To list the available tests:
197 .. code-block:: console
201 To run a single test:
203 .. code-block:: console
207 Where $TESTNAME is the name of the vsperf test you would like to run.
209 To run a group of tests, for example all tests with a name containing
212 .. code-block:: console
214 $ ./vsperf --conf-file=<path_to_custom_conf>/10_custom.conf --tests="RFC2544"
218 .. code-block:: console
220 $ ./vsperf --conf-file=<path_to_custom_conf>/10_custom.conf
222 Some tests allow for configurable parameters, including test duration
223 (in seconds) as well as packet sizes (in bytes).
227 $ ./vsperf --conf-file user_settings.py \
228 --tests RFC2544Tput \
229 --test-params "TRAFFICGEN_DURATION=10;TRAFFICGEN_PKT_SIZES=(128,)"
231 For all available options, check out the help dialog:
233 .. code-block:: console
237 Executing Vanilla OVS tests
238 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
240 1. If needed, recompile src for all OVS variants
242 .. code-block:: console
248 2. Update your ``10_custom.conf`` file to use Vanilla OVS:
250 .. code-block:: python
252 VSWITCH = 'OvsVanilla'
256 .. code-block:: console
258 $ ./vsperf --conf-file=<path_to_custom_conf>
260 Please note if you don't want to configure Vanilla OVS through the
261 configuration file, you can pass it as a CLI argument.
263 .. code-block:: console
265 $ ./vsperf --vswitch OvsVanilla
268 Executing tests with VMs
269 ^^^^^^^^^^^^^^^^^^^^^^^^
271 To run tests using vhost-user as guest access method:
273 1. Set VHOST_METHOD and VNF of your settings file to:
275 .. code-block:: python
277 VSWITCH = 'OvsDpdkVhost'
278 VNF = 'QemuDpdkVhost'
280 2. If needed, recompile src for all OVS variants
282 .. code-block:: console
290 .. code-block:: console
292 $ ./vsperf --conf-file=<path_to_custom_conf>/10_custom.conf
294 Executing tests with VMs using Vanilla OVS
295 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
297 To run tests using Vanilla OVS:
299 1. Set the following variables:
301 .. code-block:: python
303 VSWITCH = 'OvsVanilla'
304 VNF = 'QemuVirtioNet'
306 VANILLA_TGEN_PORT1_IP = n.n.n.n
307 VANILLA_TGEN_PORT1_MAC = nn:nn:nn:nn:nn:nn
309 VANILLA_TGEN_PORT2_IP = n.n.n.n
310 VANILLA_TGEN_PORT2_MAC = nn:nn:nn:nn:nn:nn
312 VANILLA_BRIDGE_IP = n.n.n.n
314 or use ``--test-params`` option
316 .. code-block:: console
318 $ ./vsperf --conf-file=<path_to_custom_conf>/10_custom.conf \
319 --test-params "VANILLA_TGEN_PORT1_IP=n.n.n.n;" \
320 "VANILLA_TGEN_PORT1_MAC=nn:nn:nn:nn:nn:nn;" \
321 "VANILLA_TGEN_PORT2_IP=n.n.n.n;" \
322 "VANILLA_TGEN_PORT2_MAC=nn:nn:nn:nn:nn:nn"
324 2. If needed, recompile src for all OVS variants
326 .. code-block:: console
334 .. code-block:: console
336 $ ./vsperf --conf-file<path_to_custom_conf>/10_custom.conf
343 Currently it is not possible to use standard scenario deployments for execution of
344 tests with VPP. It means, that deployments ``p2p``, ``pvp``, ``pvvp`` and in general any
345 :ref:`pxp-deployment` won't work with VPP. However it is possible to use VPP in
346 :ref:`step-driven-tests`. A basic set of VPP testcases covering ``phy2phy``, ``pvp``
347 and ``pvvp`` tests are already prepared.
349 List of performance tests with VPP support follows:
351 * phy2phy_tput_vpp: VPP: LTD.Throughput.RFC2544.PacketLossRatio
352 * phy2phy_cont_vpp: VPP: Phy2Phy Continuous Stream
353 * phy2phy_back2back_vpp: VPP: LTD.Throughput.RFC2544.BackToBackFrames
354 * pvp_tput_vpp: VPP: LTD.Throughput.RFC2544.PacketLossRatio
355 * pvp_cont_vpp: VPP: PVP Continuous Stream
356 * pvp_back2back_vpp: VPP: LTD.Throughput.RFC2544.BackToBackFrames
357 * pvvp_tput_vpp: VPP: LTD.Throughput.RFC2544.PacketLossRatio
358 * pvvp_cont_vpp: VPP: PVP Continuous Stream
359 * pvvp_back2back_vpp: VPP: LTD.Throughput.RFC2544.BackToBackFrames
361 In order to execute testcases with VPP it is required to:
363 * install VPP manually, see :ref:`vpp-installation`
364 * configure ``WHITELIST_NICS``, with two physical NICs connected to the traffic generator
365 * configure traffic generator, see :ref:`trafficgen-installation`
367 After that it is possible to execute VPP testcases listed above.
371 .. code-block:: console
373 $ ./vsperf --conf-file=<path_to_custom_conf> phy2phy_tput_vpp
377 Using vfio_pci with DPDK
378 ^^^^^^^^^^^^^^^^^^^^^^^^^
380 To use vfio with DPDK instead of igb_uio add into your custom configuration
381 file the following parameter:
383 .. code-block:: python
385 PATHS['dpdk']['src']['modules'] = ['uio', 'vfio-pci']
388 **NOTE:** In case, that DPDK is installed from binary package, then please
389 set ``PATHS['dpdk']['bin']['modules']`` instead.
391 **NOTE:** Please ensure that Intel VT-d is enabled in BIOS.
393 **NOTE:** Please ensure your boot/grub parameters include
396 **NOTE:** In case of VPP, it is required to explicitly define, that vfio-pci
397 DPDK driver should be used. It means to update dpdk part of VSWITCH_VPP_ARGS
398 dictionary with uio-driver section, e.g. VSWITCH_VPP_ARGS['dpdk'] = 'uio-driver vfio-pci'
400 .. code-block:: console
402 iommu=pt intel_iommu=on
404 To check that IOMMU is enabled on your platform:
406 .. code-block:: console
409 [ 0.000000] Intel-IOMMU: enabled
410 [ 0.139882] dmar: IOMMU 0: reg_base_addr fbffe000 ver 1:0 cap d2078c106f0466 ecap f020de
411 [ 0.139888] dmar: IOMMU 1: reg_base_addr ebffc000 ver 1:0 cap d2078c106f0466 ecap f020de
412 [ 0.139893] IOAPIC id 2 under DRHD base 0xfbffe000 IOMMU 0
413 [ 0.139894] IOAPIC id 0 under DRHD base 0xebffc000 IOMMU 1
414 [ 0.139895] IOAPIC id 1 under DRHD base 0xebffc000 IOMMU 1
415 [ 3.335744] IOMMU: dmar0 using Queued invalidation
416 [ 3.335746] IOMMU: dmar1 using Queued invalidation
424 To use virtual functions of NIC with SRIOV support, use extended form
425 of NIC PCI slot definition:
427 .. code-block:: python
429 WHITELIST_NICS = ['0000:05:00.0|vf0', '0000:05:00.1|vf3']
431 Where 'vf' is an indication of virtual function usage and following
432 number defines a VF to be used. In case that VF usage is detected,
433 then vswitchperf will enable SRIOV support for given card and it will
434 detect PCI slot numbers of selected VFs.
436 So in example above, one VF will be configured for NIC '0000:05:00.0'
437 and four VFs will be configured for NIC '0000:05:00.1'. Vswitchperf
438 will detect PCI addresses of selected VFs and it will use them during
441 At the end of vswitchperf execution, SRIOV support will be disabled.
443 SRIOV support is generic and it can be used in different testing scenarios.
446 * vSwitch tests with DPDK or without DPDK support to verify impact
447 of VF usage on vSwitch performance
448 * tests without vSwitch, where traffic is forwared directly
449 between VF interfaces by packet forwarder (e.g. testpmd application)
450 * tests without vSwitch, where VM accesses VF interfaces directly
451 by PCI-passthrough_ to measure raw VM throughput performance.
455 Using QEMU with PCI passthrough support
456 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
458 Raw virtual machine throughput performance can be measured by execution of PVP
459 test with direct access to NICs by PCI passthrough. To execute VM with direct
460 access to PCI devices, enable vfio-pci_. In order to use virtual functions,
461 SRIOV-support_ must be enabled.
463 Execution of test with PCI passthrough with vswitch disabled:
465 .. code-block:: console
467 $ ./vsperf --conf-file=<path_to_custom_conf>/10_custom.conf \
468 --vswitch none --vnf QemuPciPassthrough pvp_tput
470 Any of supported guest-loopback-application_ can be used inside VM with
471 PCI passthrough support.
473 Note: Qemu with PCI passthrough support can be used only with PVP test
476 .. _guest-loopback-application:
478 Selection of loopback application for tests with VMs
479 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
481 To select the loopback applications which will forward packets inside VMs,
482 the following parameter should be configured:
484 .. code-block:: python
486 GUEST_LOOPBACK = ['testpmd']
488 or use ``--test-params`` CLI argument:
490 .. code-block:: console
492 $ ./vsperf --conf-file=<path_to_custom_conf>/10_custom.conf \
493 --test-params "GUEST_LOOPBACK=['testpmd']"
495 Supported loopback applications are:
497 .. code-block:: console
499 'testpmd' - testpmd from dpdk will be built and used
500 'l2fwd' - l2fwd module provided by Huawei will be built and used
501 'linux_bridge' - linux bridge will be configured
502 'buildin' - nothing will be configured by vsperf; VM image must
503 ensure traffic forwarding between its interfaces
505 Guest loopback application must be configured, otherwise traffic
506 will not be forwarded by VM and testcases with VM related deployments
507 will fail. Guest loopback application is set to 'testpmd' by default.
509 **NOTE:** In case that only 1 or more than 2 NICs are configured for VM,
510 then 'testpmd' should be used. As it is able to forward traffic between
511 multiple VM NIC pairs.
513 **NOTE:** In case of linux_bridge, all guest NICs are connected to the same
514 bridge inside the guest.
516 Mergable Buffers Options with QEMU
517 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
519 Mergable buffers can be disabled with VSPerf within QEMU. This option can
520 increase performance significantly when not using jumbo frame sized packets.
521 By default VSPerf disables mergable buffers. If you wish to enable it you
522 can modify the setting in the a custom conf file.
524 .. code-block:: python
526 GUEST_NIC_MERGE_BUFFERS_DISABLE = [False]
528 Then execute using the custom conf file.
530 .. code-block:: console
532 $ ./vsperf --conf-file=<path_to_custom_conf>/10_custom.conf
534 Alternatively you can just pass the param during execution.
536 .. code-block:: console
538 $ ./vsperf --test-params "GUEST_NIC_MERGE_BUFFERS_DISABLE=[False]"
541 Selection of dpdk binding driver for tests with VMs
542 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
544 To select dpdk binding driver, which will specify which driver the vm NICs will
545 use for dpdk bind, the following configuration parameter should be configured:
547 .. code-block:: console
549 GUEST_DPDK_BIND_DRIVER = ['igb_uio_from_src']
551 The supported dpdk guest bind drivers are:
553 .. code-block:: console
555 'uio_pci_generic' - Use uio_pci_generic driver
556 'igb_uio_from_src' - Build and use the igb_uio driver from the dpdk src
558 'vfio_no_iommu' - Use vfio with no iommu option. This requires custom
559 guest images that support this option. The default
560 vloop image does not support this driver.
562 Note: uio_pci_generic does not support sr-iov testcases with guests attached.
563 This is because uio_pci_generic only supports legacy interrupts. In case
564 uio_pci_generic is selected with the vnf as QemuPciPassthrough it will be
565 modified to use igb_uio_from_src instead.
567 Note: vfio_no_iommu requires kernels equal to or greater than 4.5 and dpdk
568 16.04 or greater. Using this option will also taint the kernel.
570 Please refer to the dpdk documents at http://dpdk.org/doc/guides for more
571 information on these drivers.
573 Guest Core and Thread Binding
574 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
576 VSPERF provides options to achieve better performance by guest core binding and
577 guest vCPU thread binding as well. Core binding is to bind all the qemu threads.
578 Thread binding is to bind the house keeping threads to some CPU and vCPU thread to
579 some other CPU, this helps to reduce the noise from qemu house keeping threads.
582 .. code-block:: python
584 GUEST_CORE_BINDING = [('#EVAL(6+2*#VMINDEX)', '#EVAL(7+2*#VMINDEX)')]
586 **NOTE** By default the GUEST_THREAD_BINDING will be none, which means same as
587 the GUEST_CORE_BINDING, i.e. the vcpu threads are sharing the physical CPUs with
588 the house keeping threads. Better performance using vCPU thread binding can be
589 achieved by enabling affinity in the custom configuration file.
591 For example, if an environment requires 32,33 to be core binded and 29,30&31 for
592 guest thread binding to achieve better performance.
594 .. code-block:: python
596 VNF_AFFINITIZATION_ON = True
597 GUEST_CORE_BINDING = [('32','33')]
598 GUEST_THREAD_BINDING = [('29', '30', '31')]
603 QEMU default to a compatible subset of performance enhancing cpu features.
604 To pass all available host processor features to the guest.
606 .. code-block:: python
608 GUEST_CPU_OPTIONS = ['host,migratable=off']
610 **NOTE** To enhance the performance, cpu features tsc deadline timer for guest,
611 the guest PMU, the invariant TSC can be provided in the custom configuration file.
613 Multi-Queue Configuration
614 ^^^^^^^^^^^^^^^^^^^^^^^^^
616 VSPerf currently supports multi-queue with the following limitations:
618 1. Requires QEMU 2.5 or greater and any OVS version higher than 2.5. The
619 default upstream package versions installed by VSPerf satisfies this
622 2. Guest image must have ethtool utility installed if using l2fwd or linux
623 bridge inside guest for loopback.
625 3. If using OVS versions 2.5.0 or less enable old style multi-queue as shown
626 in the ''02_vswitch.conf'' file.
628 .. code-block:: python
630 OVS_OLD_STYLE_MQ = True
632 To enable multi-queue for dpdk modify the ''02_vswitch.conf'' file.
634 .. code-block:: python
636 VSWITCH_DPDK_MULTI_QUEUES = 2
638 **NOTE:** you should consider using the switch affinity to set a pmd cpu mask
639 that can optimize your performance. Consider the numa of the NIC in use if this
640 applies by checking /sys/class/net/<eth_name>/device/numa_node and setting an
641 appropriate mask to create PMD threads on the same numa node.
643 When multi-queue is enabled, each dpdk or dpdkvhostuser port that is created
644 on the switch will set the option for multiple queues. If old style multi queue
645 has been enabled a global option for multi queue will be used instead of the
648 To enable multi-queue on the guest modify the ''04_vnf.conf'' file.
650 .. code-block:: python
652 GUEST_NIC_QUEUES = [2]
654 Enabling multi-queue at the guest will add multiple queues to each NIC port when
655 qemu launches the guest.
657 In case of Vanilla OVS, multi-queue is enabled on the tuntap ports and nic
658 queues will be enabled inside the guest with ethtool. Simply enabling the
659 multi-queue on the guest is sufficient for Vanilla OVS multi-queue.
661 Testpmd should be configured to take advantage of multi-queue on the guest if
662 using DPDKVhostUser. This can be done by modifying the ''04_vnf.conf'' file.
664 .. code-block:: python
666 GUEST_TESTPMD_PARAMS = ['-l 0,1,2,3,4 -n 4 --socket-mem 512 -- '
667 '--burst=64 -i --txqflags=0xf00 '
668 '--nb-cores=4 --rxq=2 --txq=2 '
671 **NOTE:** The guest SMP cores must be configured to allow for testpmd to use the
672 optimal number of cores to take advantage of the multiple guest queues.
674 In case of using Vanilla OVS and qemu virtio-net you can increase performance
675 by binding vhost-net threads to cpus. This can be done by enabling the affinity
676 in the ''04_vnf.conf'' file. This can be done to non multi-queue enabled
677 configurations as well as there will be 2 vhost-net threads.
679 .. code-block:: python
681 VSWITCH_VHOST_NET_AFFINITIZATION = True
683 VSWITCH_VHOST_CPU_MAP = [4,5,8,11]
685 **NOTE:** This method of binding would require a custom script in a real
688 **NOTE:** For optimal performance guest SMPs and/or vhost-net threads should be
689 on the same numa as the NIC in use if possible/applicable. Testpmd should be
690 assigned at least (nb_cores +1) total cores with the cpu mask.
695 VSPERF provides options to support jumbo frame testing with a jumbo frame supported
696 NIC and traffic generator for the following vswitches:
702 3. TestPMD loopback with or without a guest
704 **NOTE:** There is currently no support for SR-IOV or VPP at this time with jumbo
707 All packet forwarding applications for pxp testing is supported.
709 To enable jumbo frame testing simply enable the option in the conf files and set the
710 maximum size that will be used.
712 .. code-block:: python
714 VSWITCH_JUMBO_FRAMES_ENABLED = True
715 VSWITCH_JUMBO_FRAMES_SIZE = 9000
717 To enable jumbo frame testing with OVSVanilla the NIC in test on the host must have
718 its mtu size changed manually using ifconfig or applicable tools:
720 .. code-block:: console
722 ifconfig eth1 mtu 9000 up
724 **NOTE:** To make the setting consistent across reboots you should reference the OS
725 documents as it differs from distribution to distribution.
727 To start a test for jumbo frames modify the conf file packet sizes or pass the option
728 through the VSPERF command line.
730 .. code-block:: python
732 TEST_PARAMS = {'TRAFFICGEN_PKT_SIZES':(2000,9000)}
734 .. code-block:: python
736 ./vsperf --test-params "TRAFFICGEN_PKT_SIZES=2000,9000"
738 It is recommended to increase the memory size for OvsDpdkVhostUser testing from the default
739 1024. Your size required may vary depending on the number of guests in your testing. 4096
740 appears to work well for most typical testing scenarios.
742 .. code-block:: python
744 DPDK_SOCKET_MEM = ['4096', '0']
746 **NOTE:** For Jumbo frames to work with DpdkVhostUser, mergable buffers will be enabled by
747 default. If testing with mergable buffers in QEMU is desired, disable Jumbo Frames and only
748 test non jumbo frame sizes. Test Jumbo Frames sizes separately to avoid this collision.
751 Executing Packet Forwarding tests
752 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
754 To select the applications which will forward packets,
755 the following parameters should be configured:
757 .. code-block:: python
762 or use ``--vswitch`` and ``--fwdapp`` CLI arguments:
764 .. code-block:: console
766 $ ./vsperf phy2phy_cont --conf-file user_settings.py \
770 Supported Packet Forwarding applications are:
772 .. code-block:: console
774 'testpmd' - testpmd from dpdk
777 1. Update your ''10_custom.conf'' file to use the appropriate variables
778 for selected Packet Forwarder:
780 .. code-block:: python
782 # testpmd configuration
784 # packet forwarding mode supported by testpmd; Please see DPDK documentation
785 # for comprehensive list of modes supported by your version.
786 # e.g. io|mac|mac_retry|macswap|flowgen|rxonly|txonly|csum|icmpecho|...
787 # Note: Option "mac_retry" has been changed to "mac retry" since DPDK v16.07
788 TESTPMD_FWD_MODE = 'csum'
789 # checksum calculation layer: ip|udp|tcp|sctp|outer-ip
790 TESTPMD_CSUM_LAYER = 'ip'
791 # checksum calculation place: hw (hardware) | sw (software)
792 TESTPMD_CSUM_CALC = 'sw'
793 # recognize tunnel headers: on|off
794 TESTPMD_CSUM_PARSE_TUNNEL = 'off'
798 .. code-block:: console
800 $ ./vsperf phy2phy_tput --conf-file <path_to_settings_py>
802 Executing Packet Forwarding tests with one guest
803 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
805 TestPMD with DPDK 16.11 or greater can be used to forward packets as a switch to a single guest using TestPMD vdev
806 option. To set this configuration the following parameters should be used.
808 .. code-block:: python
813 or use ``--vswitch`` and ``--fwdapp`` CLI arguments:
815 .. code-block:: console
817 $ ./vsperf pvp_tput --conf-file user_settings.py \
821 Guest forwarding application only supports TestPMD in this configuration.
823 .. code-block:: python
825 GUEST_LOOPBACK = ['testpmd']
827 For optimal performance one cpu per port +1 should be used for TestPMD. Also set additional params for packet forwarding
828 application to use the correct number of nb-cores.
830 .. code-block:: python
832 DPDK_SOCKET_MEM = ['1024', '0']
833 VSWITCHD_DPDK_ARGS = ['-l', '46,44,42,40,38', '-n', '4']
834 TESTPMD_ARGS = ['--nb-cores=4', '--txq=1', '--rxq=1']
836 For guest TestPMD 3 VCpus should be assigned with the following TestPMD params.
838 .. code-block:: python
840 GUEST_TESTPMD_PARAMS = ['-l 0,1,2 -n 4 --socket-mem 1024 -- '
841 '--burst=64 -i --txqflags=0xf00 '
842 '--disable-hw-vlan --nb-cores=2 --txq=1 --rxq=1']
844 Execution of TestPMD can be run with the following command line
846 .. code-block:: console
848 ./vsperf pvp_tput --vswitch=none --fwdapp=TestPMD --conf-file <path_to_settings_py>
850 **NOTE:** To achieve the best 0% loss numbers with rfc2544 throughput testing, other tunings should be applied to host
851 and guest such as tuned profiles and CPU tunings to prevent possible interrupts to worker threads.
853 VSPERF modes of operation
854 ^^^^^^^^^^^^^^^^^^^^^^^^^
856 VSPERF can be run in different modes. By default it will configure vSwitch,
857 traffic generator and VNF. However it can be used just for configuration
858 and execution of traffic generator. Another option is execution of all
859 components except traffic generator itself.
861 Mode of operation is driven by configuration parameter -m or --mode
863 .. code-block:: console
865 -m MODE, --mode MODE vsperf mode of operation;
867 "normal" - execute vSwitch, VNF and traffic generator
868 "trafficgen" - execute only traffic generator
869 "trafficgen-off" - execute vSwitch and VNF
870 "trafficgen-pause" - execute vSwitch and VNF but wait before traffic transmission
872 In case, that VSPERF is executed in "trafficgen" mode, then configuration
873 of traffic generator can be modified through ``TRAFFIC`` dictionary passed to the
874 ``--test-params`` option. It is not needed to specify all values of ``TRAFFIC``
875 dictionary. It is sufficient to specify only values, which should be changed.
876 Detailed description of ``TRAFFIC`` dictionary can be found at
877 :ref:`configuration-of-traffic-dictionary`.
879 Example of execution of VSPERF in "trafficgen" mode:
881 .. code-block:: console
883 $ ./vsperf -m trafficgen --trafficgen IxNet --conf-file vsperf.conf \
884 --test-params "TRAFFIC={'traffic_type':'rfc2544_continuous','bidir':'False','framerate':60}"
886 Code change verification by pylint
887 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
889 Every developer participating in VSPERF project should run
890 pylint before his python code is submitted for review. Project
891 specific configuration for pylint is available at 'pylint.rc'.
893 Example of manual pylint invocation:
895 .. code-block:: console
897 $ pylint --rcfile ./pylintrc ./vsperf
902 Custom image fails to boot
903 ~~~~~~~~~~~~~~~~~~~~~~~~~~
905 Using custom VM images may not boot within VSPerf pxp testing because of
906 the drive boot and shared type which could be caused by a missing scsi
907 driver inside the image. In case of issues you can try changing the drive
910 .. code-block:: python
912 GUEST_BOOT_DRIVE_TYPE = ['ide']
913 GUEST_SHARED_DRIVE_TYPE = ['ide']
915 OVS with DPDK and QEMU
916 ~~~~~~~~~~~~~~~~~~~~~~~
918 If you encounter the following error: "before (last 100 chars):
919 '-path=/dev/hugepages,share=on: unable to map backing store for
920 hugepages: Cannot allocate memory\r\n\r\n" during qemu initialization,
921 check the amount of hugepages on your system:
923 .. code-block:: console
925 $ cat /proc/meminfo | grep HugePages
928 By default the vswitchd is launched with 1Gb of memory, to change
929 this, modify --socket-mem parameter in conf/02_vswitch.conf to allocate
930 an appropriate amount of memory:
932 .. code-block:: python
934 DPDK_SOCKET_MEM = ['1024', '0']
935 VSWITCHD_DPDK_ARGS = ['-c', '0x4', '-n', '4']
936 VSWITCHD_DPDK_CONFIG = {
937 'dpdk-init' : 'true',
938 'dpdk-lcore-mask' : '0x4',
939 'dpdk-socket-mem' : '1024,0',
942 Note: Option ``VSWITCHD_DPDK_ARGS`` is used for vswitchd, which supports ``--dpdk``
943 parameter. In recent vswitchd versions, option ``VSWITCHD_DPDK_CONFIG`` will be
944 used to configure vswitchd via ``ovs-vsctl`` calls.
950 For more information and details refer to the rest of vSwitchPerfuser documentation.