1、 ETSI GS NFV-IFA 003 V2.1.1 (2016-04) Network Functions Virtualisation (NFV); Acceleration Technologies; vSwitch Benchmarking and Acceleration Specification Disclaimer The present document has been produced and approved by the Network Functions Virtualisation (NFV) ETSI Industry Specification Group
2、(ISG) and represents the views of those members who participated in this ISG. It does not necessarily represent the views of the entire ETSI membership. GROUP SPECIFICATION ETSI ETSI GS NFV-IFA 003 V2.1.1 (2016-04) 2 Reference DGS/NFV-IFA003 Keywords acceleration, benchmarking, NFV, performance, req
3、uirements, switching ETSI 650 Route des Lucioles F-06921 Sophia Antipolis Cedex - FRANCE Tel.: +33 4 92 94 42 00 Fax: +33 4 93 65 47 16 Siret N 348 623 562 00017 - NAF 742 C Association but non lucratif enregistre la Sous-Prfecture de Grasse (06) N 7803/88 Important notice The present document can b
4、e downloaded from: http:/www.etsi.org/standards-search The present document may be made available in electronic versions and/or in print. The content of any electronic and/or print versions of the present document shall not be modified without the prior written authorization of ETSI. In case of any
5、existing or perceived difference in contents between such versions and/or in print, the only prevailing document is the print of the Portable Document Format (PDF) version kept on a specific network drive within ETSI Secretariat. Users of the present document should be aware that the document may be
6、 subject to revision or change of status. Information on the current status of this and other ETSI documents is available at https:/portal.etsi.org/TB/ETSIDeliverableStatus.aspx If you find errors in the present document, please send your comment to one of the following services: https:/portal.etsi.
7、org/People/CommiteeSupportStaff.aspx Copyright Notification No part may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and microfilm except as authorized by written permission of ETSI. The content of the PDF version shall not be modified witho
8、ut the written authorization of ETSI. The copyright and the foregoing restriction extend to reproduction in all media. European Telecommunications Standards Institute 2016. All rights reserved. DECTTM, PLUGTESTSTM, UMTSTMand the ETSI logo are Trade Marks of ETSI registered for the benefit of its Mem
9、bers. 3GPPTM and LTE are Trade Marks of ETSI registered for the benefit of its Members and of the 3GPP Organizational Partners. GSM and the GSM logo are Trade Marks registered and owned by the GSM Association. ETSI ETSI GS NFV-IFA 003 V2.1.1 (2016-04) 3 Contents Intellectual Property Rights 4g3Forew
10、ord . 4g3Modal verbs terminology 4g31 Scope 5g32 References 5g32.1 Normative references . 5g32.2 Informative references 5g33 Definitions and abbreviations . 6g33.1 Definitions 6g33.2 Abbreviations . 6g34 Overview 7g34.1 Problem Statement . 7g34.2 vSwitch Use Cases . 8g34.2.1 Virtual Forwarding . 8g3
11、4.2.2 Overlay based Virtual Networks . 8g34.2.3 Traffic Filtering 9g34.2.4 Distributed Network Services . 9g34.2.5 Traffic Monitoring 9g34.2.6 Load Balancing . 10g34.2.7 Latency/Jitter Sensitive Workloads 10g34.2.8 Efficient Policy and QoS Control . 10g34.2.9 Traffic Control Essential, or potentiall
12、y Essential, IPRs notified to ETSI in respect of ETSI standards“, which is available from the ETSI Secretariat. Latest updates are available on the ETSI Web server (https:/ipr.etsi.org/). Pursuant to the ETSI IPR Policy, no investigation, including IPR searches, has been carried out by ETSI. No guar
13、antee can be given as to the existence of other IPRs not referenced in ETSI SR 000 314 (or the updates on the ETSI Web server) which are, or may be, or may become, essential to the present document. Foreword This Group Specification (GS) has been produced by ETSI Industry Specification Group (ISG) N
14、etwork Functions Virtualisation (NFV). Modal verbs terminology In the present document “shall“, “shall not“, “should“, “should not“, “may“, “need not“, “will“, “will not“, “can“ and “cannot“ are to be interpreted as described in clause 3.2 of the ETSI Drafting Rules (Verbal forms for the expression
15、of provisions). “must“ and “must not“ are NOT allowed in ETSI deliverables except when used in direct citation. ETSI ETSI GS NFV-IFA 003 V2.1.1 (2016-04) 5 1 Scope The present document specifies performance benchmarking metrics for virtual switching, with the goal that the metrics will adequately qu
16、antify performance gains achieved through virtual switch acceleration conforming to the associated requirements specified herein. The acceleration-related requirements will be applicable to common virtual switching functions across usage models such as packet delivery into VNFs, network overlay and
17、tunnel termination, stateful Network Address Translators (NAT), service chaining, load balancing and, in general, match-action based policies/flows applied to traffic going to/from the VMs. The present document will also provide deployment scenarios with applicability to multiple vendor implementati
18、ons and recommendations for follow-on proof of concept activities. 2 References 2.1 Normative references References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For specific references, only the cited version applies. For non-specif
19、ic references, the latest version of the referenced document (including any amendments) applies. Referenced documents which are not found to be publicly available in the expected location might be found at http:/docbox.etsi.org/Reference. NOTE: While any hyperlinks included in this clause were valid
20、 at the time of publication, ETSI cannot guarantee their long term validity. The following referenced documents are necessary for the application of the present document. 1 ETSI GS NFV 003: “Network Functions Virtualisation (NFV); Terminology for main concepts in NFV“. 2 IETF RFC 2544: “Benchmarking
21、 Methodology for Network Interconnect Devices“. 3 IETF RFC 2679: “A One-way Delay Metric for IPPM“. 4 IETF RFC 2680: “A One-way Loss Metric for IPPM“. 5 IETF RFC 3511: “Benchmarking Methodology for Firewall Performance“. 6 IETF RFC 4737: “Packet Reordering Metrics“. 7 IETF RFC 5481: “Packet Delay Va
22、riation Applicability Statement“. 8 IETF RFC 6703: “Reporting IP Network Performance Metrics: Different Points of View“. 2.2 Informative references References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For specific references, onl
23、y the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies. NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee their long term validity. The following referenc
24、ed documents are not necessary for the application of the present document but they assist the user with regard to a particular subject area. i.1 IETF: draft-ietf-bmwg-ipsec-term-12.txt, “Terminology for Benchmarking IPsec Devices“. i.2 IETF: draft-ietf-bmwg-ipsec-meth-05.txt, “Methodology for Bench
25、marking IPsec Devices“. i.3 IETF: draft-ietf-bmwg-virtual-net-01.txt, “Considerations for Benchmarking Virtual Network Functions and Their Infrastructure“. i.4 IETF: draft-vsperf-bmwg-vswitch-opnfv-01.txt, “Benchmarking Virtual Switches in OPNFV“. ETSI ETSI GS NFV-IFA 003 V2.1.1 (2016-04) 6 i.5 IETF
26、 RFC 6049: “Spatial Composition of Metrics“. i.6 IETF RFC 7348: “Virtual eXtensible Local Area Network (VXLAN): A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks“. i.7 ETSI GS NFV-INF 007: “ETSI GS NFV INF-007:“ Network Functions Virtualisation (NFV); Management and Orche
27、stration; Or-Vnfm reference point - Interface and Information Model Specification“. i.8 ETSI GS NFV-IFA 002: “Network Functions Virtualisation (NFV); Acceleration Technologies; VNF Interfaces Specifications“. i.9 IETF RFC 6815: “Applicability Statement for RFC 2544: Use on Production Networks Consid
28、ered Harmful“. i.10 IETF RFC 6985: “IMIX Genome: Specification of Variable Packet Sizes for Additional Testing“. 3 Definitions and abbreviations 3.1 Definitions For the purposes of the present document, the terms and definitions given in ETSI GS NFV 003 1 apply. 3.2 Abbreviations For the purposes of
29、 the present document, the following abbreviations apply: ACK Acknowledge ACPI Advanced Configuration and Power Interface ARP Address Resolution Protocol BIOS Basic Input Output System BS Block Storage CPU Central Processing Unit DIMM Dual In-line Memory Module DPDK Data Plane Development Kit DUT De
30、vice Under Test GRUB Grand Unified Bootloader HTTP HyperText Transfer Protocol IMIX Internet Mix IO Input Output IP Internet ProtocolMAC Media Access Control MTU Maximum Transmission Unit NAT Network Address Translation NFP Network Forwarding Path NFV Network Functions Virtualisation NFVI NFV Infras
31、tructure NIC Network Interface Card NUMA Non Uniform Memory Access OPNFV Open Platform for NFV PCI Peripheral Component Interconnect PDV Packet Delay Variation RSS Receive Side Scaling SF Service Function SFC Service Function Chaining SFF Service Function Forwarders SLA Service Level Agreement SUT S
32、ystem Under Test SW Software TCP Transmission control Protocol TSO TCP Segment Offload ETSI ETSI GS NFV-IFA 003 V2.1.1 (2016-04) 7 UML Unified Modelling Language VBS Virtual Block Storage VIM Virtual Infrastructure Manager VLAN Virtual eXtensible Local Area Network VM Virtual Machine VNF Virtualised
33、 Network Function VNFC Virtual Network Function Component VNFD Virtual Network Function Description VNFFG VNF Forwarding Graph VNI VxLAN Network Identifier VSPERF OPNFV vSwitch Performance Project VXLAN Virtual eXtensible Local Area Network 4 Overview 4.1 Problem Statement Inside a compute platform
34、a virtual switch (vSwitch) is used to interconnect VNFs that share the same platform. The vSwitch is in a unique position of being at the intersection of the network and the VNFs themselves. As such the implementation and specifics of the virtual switching on the platform need to be transparent to t
35、he VNFs in the system, thus as opposed to ETSI GS NFV-IFA 002 i.8 where VNFs are requesting acceleration, vSwitch acceleration needs to be transparent to individual VNFs, and controlled from the VIM. NOTE: In the context of the present document, vSwitch may include some of the functionality of vRout
36、er as defined in ETSI GS NFV-INF 007 i.7. A flow within the vSwitch is given as the classification (locator + domain) and the port forwarding action. This definition of flow differs from what is tracked as a flow within a VNF or even within the VIM. In this context the vSwitch is only concerned with
37、 the flow information needed to perform the virtual switching functionality. Since the vSwitch finds itself in this unique position between the rest of the network and the VNFs, it is very common to add additional functionality at this point of control. In the present document, these are called in-l
38、ine functions, and are defined as network services that have been placed in-line with the switching function. Examples of in-line functions include: ACLs: Doing a more complex (usually wildcard based) classification for security and monitoring purposes Tunnel Endpoint: Pushing packets in and out of
39、a tunnel in order to traverse a physical network Address Translation/NAT: Translating packet headers to expand the address space of the network Load Balancing: Choosing from a set of destinations to forward a packet QoS: assigning a class of service for the purpose of traffic shaping, rate limiting,
40、 priority queuing, mapping of per-packet features of VNF to infrastructure These in-line functions are logically separate from the baseline virtual switching function, and as such may have their own specific definition of what constitutes a flow, and what additional classification and state informat
41、ion is tracked. In addition to above in-line functions, stateful operations such as Firewall or Load Balancer may be implemented. In contrast to in-line functions, network functions could also sit within a VNF, in which case a vSwitch may choose to classify the packet and switch it to this VNF. In o
42、rder to provide service-to-service context and to preserve the initial classification of the packet, service chaining may be used to position more complicated functions inside or outside of a VNF. ETSI ETSI GS NFV-IFA 003 V2.1.1 (2016-04) 8 The present document defines the critical aspects of vSwitc
43、h performance by treating the vSwitch as a Device Under Test (DUT), with specific configurations that are consistent across instantiations of a vSwitch on a compute platform. Existing testing and benchmarks specifications (see i.1, i.2, i.3 and i.4) should be used to measure the performance of the D
44、UT under these configurations and conditions, including measurement of metrics that support service engineering (such as the Composition Functions defined in IETF RFC 6049 i.5). The following configurations are of importance (see clause 7 for more detail and diagrams): vSwitch Physical to Physical:
45、A vSwitch configured to receive traffic from a physical interface (uplink), make a forwarding decision, and re-forward the frame back out a physical interface (uplink). vSwitch Virtual to Virtual: A vSwitch configured to receive traffic from a VNF, make a forwarding decision, and re-forward the fram
46、e back out to a VNF. vSwitch VNF Loopback: A vSwitch is configured to receive traffic from a physical interface (uplink), make a forwarding decision, and then forward the frame to a VNF. The VNF should simply loopback the frames back to the vSwitch, which should do another forwarding decision to pus
47、h the frame back out a physical interface (VNF). In each configuration, the vSwitch may have a specific set of in-line functions configured such as L2 forwarding rules, L3 forwarding rules, tunnel termination, and wildcard rules used for ACLs, QoS, and monitoring. These in-line functions define the
48、use case under test. 4.2 vSwitch Use Cases 4.2.1 Virtual Forwarding The function of a vSwitch is minimally defined as a classification based on a locator (derived from the packet header) and a domain, both of which are matched upon to deliver the packet to a destination. The domain is derived either
49、 from the packet header (for example a VLAN, or VNI) or from the ingress port on the vSwitch. The packet destination is either a port on the vSwitch, or a logical port that pushes the frames into a tunnel to send it to another vSwitch across the physical network. A vSwitch also needs to correctly handle broadcast, such that protocols such as ARP are correctly propagated between VNFs. Lastly, any packets not associated with current classification rules need to be handled in a specified default manner,
