| Internet Engineering Task Force | P. Ashwood-Smith |
| Internet-Draft | Huawei Technologies |
| Intended status: Informational | R. Iyengar |
| Expires: January 15, 2014 | T. Tsou |
| Huawei Technologies USA | |
| A. Sajassi | |
| Cisco Technologies | |
| M. Boucadair | |
| C. Jacquenet | |
| France Telecom | |
| M. Daikoku | |
| KDDI corporation | |
| July 14, 2013 |
NVO3 Operational Requirements
draft-ashwood-nvo3-operational-requirement-03
This document provides framework and requirements for Network Virtualization over Layer 3 (NVO3) Operations, Administration, and Maintenance (OAM). This document for the most part gathers requirements from existing IETF drafts and RFCs which have already extensively studied this subject for different data planes and layering. As a result this draft is high level and broad. We begin to ask which are truly required for NVO3 and expect the list to be narrowed by the working group as subsequent versions of this draft are created.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on January 15, 2014.
Copyright (c) 2013 IETF Trust and the persons identified as the document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
This document provides framework and requirements for Network virtualization over Layer 3(NVO3) Operation, Administration, and Maintenance (OAM). Given that this OAM subject is far from new and has been under extensive investigation by various IETF working groups (and several other standards bodies) for many years, this document draws from existing work, starting with [RFC6136]. As a result, sections of [RFC6136] have been reused with minor changes with the permission of the authors.
NVO3 OAM requirements are expected to be a subset of IETF/IEEE etc. work done so far; however, we begin with a full set of requirements and expect to prune them through several iterations of this document.
The scope of OAM for any service and/or transport/network infrastructure technologies can be very broad in nature. OSI has defined the following five generic functional areas commonly abbreviated as "FCAPS" [NM-Standards]:
This document focuses on the Fault, Performance and to a limited extent the Configuration Management aspects. Other functional aspects of FCAPS and their relevance (or not) to NVO3 are for further study.
Fault Management can typically be viewed in terms of the following categories:
Fault detection deals with mechanism(s) that can detect both hard failures such as link and device failures, and soft failures, such as software failure, memory corruption, misconfiguration, etc. Fault detection relies upon a set of mechanisms that first allow the observation of an event, then the use of a protocol to dynamically notify a network/system operator (or management system) about the event occurrence, then the use of diagnostic tools to assess the nature and severity of the fault.
After verifying that a fault has occurred along the data path, it is important to be able to isolate the fault to the level of a given device or link. Therefore, a fault isolation mechanism is needed in Fault Management. A fault notification mechanism should be used in conjunction with a fault detection mechanism to notify the devices upstream and downstream to the fault detection point. The fault notification mechanism should also notify NMS systems.
For example, when there is a client/server relationship between two layered networks (e.g., the NVO3 layer is a client of the outer IP server layer, while the inner IP layer is a client of the NVO3 server layer 2), fault detection at the server layer may result in the following fault notifications:
Finally, fault recovery deals with recovering from the detected failure by switching to an alternate available data path (depending on the nature of the fault) using alternate devices or links. In fact, the controller can provision another virtual network, thus automatically resolving the reported problem.
The controller may also directly monitor the status of virtual network components such as Network Virtualization Edge elements (NVEs) [NVO3-framework] in order to respond to their failures. In addition to forward and backward fault notifications, the controller may deliver notifications to a higher level orchestration component, e.g., one responsible for Virtual Machine (VM) provisioning and management.
Note, given that the IP network on which NVO3 resides is usually self healing, it is expected that recovery by the NVO3 layer would not normally be required, although there may be a requirement for that layer to log that the problem has been detected and resolved. The special cases of a static IP overlay network, or possibly of a centrally controlled IP overlay network, may, however, require NVO3 involvement in fault recovery.
Performance Management deals with mechanism(s) that allow determining and measuring the performance of the network/services under consideration. Performance Management can be used to verify the compliance to both the service-level and network-level metric objectives/specifications. Performance Management typically consists of measuring performance metrics, e.g., Frame Loss, Frame Delay, Frame Delay Variation (aka Jitter), Frame throughput, Frame discard, etc., across managed entities when the managed entities are in available state. Performance Management is suspended across unavailable managed entities.
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].
This document leverages requirements that originate with other OAM work, specifically the following:
The terminology defined in [NVO3-framework] and [NVO3-DP-Reqs] is used throughout this document. We introduce no new terminology.
Figure 1 below reproduces the generic NVO3 reference model as per [NVO3-framework].
+--------+ +--------+
| Tenant | | Tenant |
| End +--+ +---| End |
| System | | | | System |
+--------+ | ................... | +--------+
| +-+--+ +--+-+ |
| | NV | | NV | |
+--|Edge| |Edge|--+
+-+--+ +--+-+
/ . L3 Overlay . \
+--------+ / . Network . \ +--------+
| Tenant +--+ . . +----| Tenant |
| End | . . | End |
| System | . +----+ . | System |
+--------+ .....| NV |........ +--------+
|Edge|
+----+
|
|
+--------+
| Tenant |
| End |
| System |
+--------+
Figure 1: Generic reference model for DC network virtualization over a Layer3 infrastructure
Figure 2 below, reproduces the Generic reference model for the NV Edge (NVE) as per [NVO3-DP-Reqs].
+------- L3 Network ------+
| |
| Tunnel Overlay |
+------------+---------^-+ +--------+-------------^-+
| +----------+------+ | | | +------+----------+ | |
| | Overlay Module | | | | | Overlay Module | | |
| +--------+--------+ | | | +--------+--------+ | |
| | VNID | | | | VNID | |
| | OAM | | | OAM |
| +-------+-------+ | | | +-------+-------+ | |
| | VNI | | | | | VNI | | |
NVE1 | +-+-----------+-+ | | NVE2 | +-+-----------+-+ | |
| | VAPs | | | | | VAPs | | |
+----+-----------+-----V-+ +----+-----------+-----V-+
| | | |
-------+-----------+--------------------+-----------+-----_--
| | Tenant | |
| | Service IF | |
Tenant End Systems Tenant End Systems
Figure 2: Generic reference model for NV Edge
Figure 1 showed the generic reference model for a DC network virtualization over an L3 (or L3VPN) infrastructure while Figure 2 showed the generic reference model for the Network Virtualization (NV) Edge.
L3 network(s) or L3 VPN networks (either IPv6 or IPv4, or a combination thereof), provide transport for an emulated layer 2 created by NV Edge devices. Unicast and multicast tunneling methods (de-multiplexed by Virtual Network Identifier (VNID)) are used to provide connectivity between the NV Edge devices. The NV Edge devices then present an emulated layer 2 network to the Tenant End Systems at a Virtual Network Interface (VNI) through Virtual Access Points (VAPs). The NV Edge devices map layer 2 unicast to layer 3 unicast point-to-point tunnels and may either map layer 2 multicast to layer 3 multicast tunnels or may replicate packets onto multiple layer 3 unicast tunnels.
The emulated layer 2 network is provided by the NV Edge devices to which the Tenant End Systems are connected. This network of NV Edges can be operated by a single service provider or can span across multiple administrative domains. Likewise, the L3 Overlay Network can be operated by a single service provider or span across multiple administrative domains.
While each of the layers is responsible for its own OAM, each layer may consist of several different administrative domains. Figure 3 shows an example.
OAM
---
TENANT |----------------------------| TENANT {all IP/ETH}
NV Edge |----------------------| NV Edge {t.b.d.}
IP(VPN) |---| IP (VPN) |---| IP(VPN) {IP(VPN)/ETH}
Figure 3: OAM layers in an NVO3 network
For example, at the bottom, at the L3 IP overlay network layer IP(VPN) and/or Ethernet OAM mechanisms are used to probe link by link, node to node etc. OAM addressing here means physical node loopback or interface addresses.
Further up, at the NV Edge layer, NVO3 OAM messages are used to probe the NV Edge to NV Edge tunnels and NV Edge entity status. OAM addressing here likely means the physical node loopback together with the VNI (to de-multiplex the tunnels).
Finally, at the Tenant layer, the IP and/or Ethernet OAM mechanisms are again used but here they are operating over the logical L2/L3 provided by the NV-Edge through the VAP. OAM addressing at this layer deals with the logical interfaces on Vswitches and Virtual Machines.
Complex OAM relationships exist as a result of the hierarchical layering of responsibility and of breaking up of end-to-end responsibility.
The OAM domain above NVO3, is expected to be supported by existing IP and L2 OAM methods and tools.
The OAM domain below NVO3, is expected to be supported by existing IP/L2 and MPLS OAM methods and tools. Where this layer is actually multiple domains spliced together, the existing methods to deal with these boundaries are unchanged. Note however that exposing LAG/ECMP detailed behavior may result in additional requirements to this domain, the details of which will be specified in the future versions of this draft.
When we refer to an OAM domain in this document, or just 'domain', we therefore refer to a closed set of NV Edges and the tunnels which interconnect them. Inter-domain OAM considerations will be specified in the future versions of this draft.
The following numbered requirements originate from [RFC6136]. All are included however where they seem obviously not relevant (to the present authors) an explanation as to why is included.
R1) NVO3 OAM MUST allow an NV Edge device to dynamically discover other NV Edge devices that share the same VNI within a given NVO3 domain. This may be based on a discovery mechanism used to set up data path forwarding between NVEs.
R2) NVO3 OAM MUST allow proactive connectivity monitoring between two or more NV Edge devices that support the same VNIs within a given NVO3 domain. NVO3 OAM MAY act as a protection trigger. That is, automatic recovery from transmission facility failure by switchover to a redundant replacement facility may be triggered by notifications from NVO3 OAM.
R3) NVO3 OAM MUST allow monitoring/tracing of all possible paths in the underlay network between a specified set of two or more NV Edge devices. Using this feature, equal cost paths that traverse LAG and/or ECMP may be differentiated.
R4) NVO3 OAM MUST allow connectivity fault verification between two or more NV Edge devices that support the same VNI within a given NVO3 domain.
R5) NVO3 OAM MUST allow connectivity fault localization between two or more NV Edge devices that support the same VNI within a given NVO3 domain.
R6) NVO3 OAM MUST support fault notification to be triggered as a result of the faults occurring in the underneath network infrastructure. This fault notification SHOULD be used for the suppression of redundant service-level alarms.
R7) NVO3 OAM MUST support measurement of per VNI frame loss between two NV Edge devices that support the same VNI within a given NVO3 domain.
R8) NVO3 OAM MUST support measurement of per VNI two-way frame delay between two NV edge devices that support the same VNI within a given NVO3 domain.
R9) NVO3 OAM MUST support measurement of per VNI one-way frame delay between two NV Edge devices that support the same VNI within a given NVO3 domain.
R10) NVO3 OAM MUST support measurement of per VNI frame delay variation between two NV Edge devices that support the same VNI within a given NVO3 domain.
R11) NVO3 OAM MAY [*** Should this be stronger? ***] support measurement of per VNI frame throughput throughput (in frames and bytes) between two NV Edge devices that support the same VNI within a given NVO3 domain. This feature could be an effective way to confirm whether or not assigned path bandwidth conforms to service level agreement before providing the path between two NV Edge devices.
R12) NVO3 OAM MAY support measurement of per VNI frame discard between two NV Edge devices that support the same VNI within a given NVO3 domain. This feature MAY be effective to monitor bursty traffic between two NV Edge devices.
A service may be considered unavailable if the service frames/packets do not reach their intended destination (e.g., connectivity is down) or the service is degraded (e.g., frame loss and/or frame delay and/or delay variation threshold is exceeded). Entry and exit conditions may be defined for the unavailable state. Availability itself may be defined in the context of a service type. Since availability measurement may be associated with connectivity, frame loss, frame delay, and frame delay variation measurements, no additional requirements are specified currently.
R13) NVO3 OAM frames MUST be forwarded along the same path (i.e., links (including LAG members) and nodes) as the NVO3 data frames.
R14) NVO3 OAM frames MUST provide a mechanism to exercise/trace all data paths that result due to ECMP/LAG hops in the underlay network.
R15) NVO3 OAM MUST be scalable such that an NV edge device can support proactive OAM for each VNI that is supported by the device. (Note - Likely very hard to achieve with hash based ECMP/LAG).
R16) NVO3 OAM should be extensible such that new functionality and information elements related to this functionality can be introduced in the future.
R17) NVO3 OAM MUST be defined such that devices not supporting the OAM are able to forward the OAM frames in a similar fashion as the regular NVO3 data frames/packets.
R18) NVO3 OAM frames MUST be prevented from leaking outside their NVO3 domain.
R19) NVO3 OAM frames from outside an NVO3 domain MUST be prevented from entering the said NVO3 domain when such OAM frames belong to the same level or to a lower-level OAM. (Trivially met because hierarchical domains are independent technologies.)
R20) NVO3 OAM frames from outside an NVO3 domain MUST be transported transparently inside the NVO3 domain when such OAM frames belong to a higher-level NVO3 domain. (Trivially met because hierarchical domains are independent technologies).
Similar to transport requirement from [RFC6136], we expect NVO3 OAM will leverage the OAM capabilities of the transport layer (e.g., IP underlay).
R21) NVO3 OAM MAY allow adaptation/interworking with its IP underlay OAM functions. For example, this would be useful to allow fault notifications from the IP layer to be sent to the NVO3 layer and likewise exposure of LAG / ECMP will require such non-independence.
R22) NVO3 OAM MUST [*** discuss -- is this too strong? ***] be independent of the application technologies and specific application OAM capabilities.
[Comment -- ECM: Noticed Nicira implementation has a dedicated NVP manager node to play the role of FCAPS here. It is both application layer and OAM layer. May not meet this requirement. In reality, due to the nature of overlay network, very often, vendors are going to make everything all together to a dedicated manager node.]
R23) NVO3 OAM messages MUST be preferentially treated in NVE and between NVEs, since NVO3 OAM MAY be used to trigger protection switching. As noted above (R2), protection switching is the automatic replacement of a failed transmission facility with a working one providing equal or greater capacity, typically within a few tens of milliseconds from fault detection.
[Comment -- ECM: giving NVO3 OAM messages priority treatment may interfere with measurements of frame delay and jitter.]
This section identifies a set of operational items which may be elaborated further if these items fall within the scope of the NVO3.
This memo includes no request to IANA.
TBD
The authors are grateful for the contributions of David Black, Dennis Qin, Erik Smith and Ziye Yang to this latest version.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. |
| [RFC6136] | Sajassi, A. and D. Mohan, "Layer 2 Virtual Private Network (L2VPN) Operations, Administration, and Maintenance (OAM) Requirements and Framework", RFC 6136, March 2011. |
| [NVO3-framework] | Lasserre, M., Balus, F., Morin, T., Bitar, N. and Y. Rekhter, "Framework for DC Network Virtualization", July 2012. |
| [NVO3-DP-Reqs] | Bitar, N., Lasserre, M., Balus, F., Morin, T., Jin, L. and B. Khasnabish, "NVO3 Data Plane Requirements", October 2012. |
| [IEEE802.1ag] | , , "IEEE Standard for Local and metropolitan area networks - Virtual Bridged Local Area Networks, Amendment 5: Connectivity Fault Management", 2007. |
| [IEEE802.1ah] | , , "IEEE Standard for Local and metropolitan area networks - Virtual Bridged Local Area Networks, Amendment 6: Provider Backbone Bridges", 2008. |
| [Y.1731] | , , "ITU-T Recommendation Y.1731 (02/08) - OAM functions and mechanisms for Ethernet based networks", February 2008. |
| [NM-Standards] | , , "ITU-T Recommendation M.3400 (02/2000) - TMN Management Functions", February 2000. |