L2VPN Workgroup J. Rabadan Internet Draft S. Sathappan Intended status: Standards Track W. Henderickx S. Palislamovic Alcatel-Lucent F. Balus Nuage Networks Expires: August 18, 2014 February 14, 2014 Data Center Interconnect Solution for EVPN Overlay networks draft-rabadan-l2vpn-dci-evpn-overlay-01.txt Abstract This document describes how Network Virtualization Overlay networks (NVO3) can be connected to a Wide Area Network (WAN) in order to extend the layer-2 connectivity required for some tenants. The solution will analyze the interaction between NVO3 networks running EVPN and other L2VPN technologies used in the WAN, such as VPLS/PBB- VPLS or EVPN/PBB-EVPN, and will propose a solution for the interworking between both. Status of this Memo 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), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. 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." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html Rabadan et al. Expires August 18, 2014 [Page 1] Internet-Draft DCI for E-VPN Overlays February 14, 2014 This Internet-Draft will expire on August 18, 2014. Copyright Notice Copyright (c) 2014 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. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Decoupled DCI solution for EVPN overlay networks . . . . . . . 3 2.1. Interconnect requirements . . . . . . . . . . . . . . . . . 4 2.2. VLAN-based hand-off . . . . . . . . . . . . . . . . . . . . 5 2.3. Pseudowire-based hand-off . . . . . . . . . . . . . . . . . 5 2.4. Multi-homing solution . . . . . . . . . . . . . . . . . . . 6 2.5. Data Center Gateway Optimizations . . . . . . . . . . . . . 7 2.5.1 Use of the Unknown MAC route to reduce unknown flooding . . . . . . . . . . . . . . . . . . . . . . . . 7 2.5.2. MAC address advertisement control . . . . . . . . . . . 7 2.5.3. ARP flooding control . . . . . . . . . . . . . . . . . 8 3. Integrated DCI solution for EVPN overlay networks . . . . . . . 8 3.1. Interconnect requirements . . . . . . . . . . . . . . . . . 9 3.2. VPLS DCI for EVPN-Overlay networks . . . . . . . . . . . . 10 3.2.1. Control/Data Plane setup procedures on the DC GWs . . . 10 3.2.2. Multi-homing procedures on the DC GWs . . . . . . . . . 11 3.3. PBB-VPLS DCI for EVPN-Overlay networks . . . . . . . . . . 11 3.3.1. Control/Data Plane setup procedures on the DC GWs . . . 11 3.3.2. Multi-homing procedures on the DC GWs . . . . . . . . . 12 3.4. EVPN-MPLS DCI for EVPN-Overlay networks . . . . . . . . . . 12 3.4.1. Control Plane setup procedures on the DC GWs . . . . . 12 3.4.2. Data Plane setup procedures on the DC GWs . . . . . . . 14 3.4.3. Multi-homing procedures on the DC GWs . . . . . . . . . 14 3.4.4. Impact on MAC Mobility procedures . . . . . . . . . . . 15 3.4.5. Data Center Gateway optimizations . . . . . . . . . . . 16 3.4.6. Benefits of the EVPN-MPLS DCI solution . . . . . . . . 16 3.5. PBB-EVPN DCI for EVPN-Overlay networks . . . . . . . . . . 17 3.5.1. Control/Data Plane setup procedures on the DC GWs . . . 17 Rabadan et al. Expires August 18, 2014 [Page 2] Internet-Draft DCI for E-VPN Overlays February 14, 2014 3.5.2. Multi-homing procedures on the DC GWs . . . . . . . . . 18 3.5.3. Impact on MAC Mobility procedures . . . . . . . . . . . 18 3.5.4. Data Center Gateway optimizations . . . . . . . . . . . 18 5. Conventions and Terminology . . . . . . . . . . . . . . . . . . 18 6. Security Considerations . . . . . . . . . . . . . . . . . . . . 19 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . . 19 8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 19 8.1. Normative References . . . . . . . . . . . . . . . . . . . 19 8.2. Informative References . . . . . . . . . . . . . . . . . . 19 9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . 20 10. Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . 20 1. Introduction [EVPN-Overlays] discusses the use of EVPN as the control plane for Network Virtualization Overlay (NVO) networks, where VXLAN, NVGRE or MPLS over GRE can be used as possible data plane encapsulation options. While this model provides a scalable and efficient multi-tenant solution within the Data Center, it might not be easily extended to the WAN in some cases due to the requirements and existing deployed technologies. For instance, a Service Provider might have an already deployed (PBB-)VPLS or (PBB-)EVPN network that must be used to interconnect Data Centers and WAN VPN users. This document describes a Data Center Interconnect (DCI) solution for E-VPN overlay Data Center networks, assuming that the Data Center Gateway (DC GW) and the WAN Edge functions can be decoupled in two separate systems or integrated into the same system. The former option will be referred as "Decoupled DCI solution" throughout the document whereas the latter one will be referred as "Integrated DCI solution". 2. Decoupled DCI solution for EVPN overlay networks This section describes the interconnect solution when the DC GW and WAN Edge functions implemented in different systems. Figure 1 depicts the reference model described in this section. Rabadan et al. Expires August 18, 2014 [Page 3] Internet-Draft DCI for E-VPN Overlays February 14, 2014 +--+ |CE| +--+ | +----+ +----| PE |----+ +---------+ | +----+ | +---------+ +----+ | +---+ +----+ +----+ +---+ | +----+ |NVE1|--| |DC | |WAN | |WAN | |DC | |--|NVE3| +----+ | |GW1|--|Edge| |Edge|--|GW3| | +----+ | +---+ +----+ +----+ +---+ | | DC1 | | WAN | | DC2 | | +---+ +----+ +----+ +---+ | | |DC | |WAN | |WAN | |DC | | +----+ | |GW2|--|Edge| |Edge|--|GW4| | +----+ |NVE2|--| +---+ +----+ +----+ +---+ |--|NVE4| +----+ +---------+ | | +---------+ +----+ +--------------+ |<-EVPN-Overlay-->|<-VLAN->|<-WAN L2VPN->|<--PW-->|<--EVPN-Overlay->| hand-off hand-off Figure 1 Decouple DCI model The following section describes the interconnect requirements that make Service Providers select this model and the requirements of the solution itself. 2.1. Interconnect requirements The proposed Interconnect architecture will be normally deployed in networks where the EVPN-Overlay provider and WAN providers are different entities and a clear demarcation is needed. The solution must observe the following requirements: o A simple connectivity hand-off must be provided between the EVPN- Overlay network provider and the WAN provider so that QoS and security enforcement are easily accomplished. o The solution must be independent of the L2VPN technology deployed in the WAN. o Multi-homing between DC GW and WAN Edge routers is required. Per- service load balancing MUST be supported. Per-flow load balancing MAY be supported but it is not a strong requirement since a deterministic path per service is usually required for an easy QoS and security enforcement. Rabadan et al. Expires August 18, 2014 [Page 4] Internet-Draft DCI for E-VPN Overlays February 14, 2014 o Ethernet OAM and Connectivity Fault Management (CFM) functions must be supported between the EVPN-Overlay network and the WAN network. o The following optimizations MAY be supported at the DC GW: + Unknown flooding reduction for the unicast traffic sourced from the DC Network Virtualization Edge devices (NVEs). + Control of the WAN MAC addresses advertised to the DC. + ARP flooding control for the requests coming from the WAN. 2.2. VLAN-based hand-off In this option, the hand-off between the DC GWs and the WAN Edge routers is based on 802.1Q VLANs. This is illustrated in Figure 1, between the DC GWs in DC1 and the WAN Edge routers. Each EVPN Instance (EVI) in the DC GW is connected to a different VPLS/EVI instance in the WAN Edge router by using a different C-TAG VLAN ID or a different combination of S-TAG/C-TAG VLAN IDs that matches at both sides. In this use-case, the WAN Edge router becomes a VPLS/EVPN PE with regular Attachment Circuits. This option provides the best possible demarcation between the DC and WAN providers and it does not require control plane interaction between both providers. The disadvantages of this model are the provisioning overhead and the reduced scalability (limited to the VLAN-ID space). In this model, the DC GW acts as a regular Network Virtualization Edge (NVE) towards the DC. Its control plane, data plane procedures and interactions are described in [EVPN-Overlays]. The WAN Edge router acts as a (PBB-)VPLS or (PBB-)EVPN PE. Its functions are described in [RFC4761][RFC4762][RFC6074] or [EVPN][PBB- EVPN]. 2.3. Pseudowire-based hand-off If MPLS can be enabled between the DC GW and the WAN Edge router, a more scalable DCI solution can be deployed. In this option the hand- off between both routers is based on FEC128-based pseudowires or, alternatively, FEC129-based pseudowires for a greater level of network automation. Note that this model still provides a clear demarcation boundary between DC and WAN, and security/QoS policies may be applied on a per pseudowire basis. The PW-based hand-off interconnect is illustrated in Figure 1, between the DC2 DC GWs and the WAN Edge routers. In this model, besides the usual MPLS procedures between DC GW and WAN Edge router, the DC GW MUST support an interworking function in Rabadan et al. Expires August 18, 2014 [Page 5] Internet-Draft DCI for E-VPN Overlays February 14, 2014 each EVI that requires extension to the WAN: o If a FEC128-based pseudowire is used between the EVI (DC GW) and the VSI (WAN Edge), the provisioning of the VCID for such pseudowire MUST be supported on the EVI and must match the VCID used in the peer VSI at the WAN Edge router. o If BGP Auto-discovery [RFC6074] and FEC129-based pseudowires are used between the DC GW EVI and the WAN Edge VSI, the provisioning of the VPLS-ID MUST be supported on the EVI and must match the VPLS-ID used in the WAN Edge VSI. 2.4. Multi-homing solution As already discussed, single-active multi-homing, i.e. per-service load-balancing multi-homing MUST be supported in this type of interconnect. All-active multi-homing may be considered in future revisions of this document. The DC GWs will be provisioned with a unique ESI per WAN interconnect and the hand-off attachment circuits or pseudowires between the DC GW and the WAN Edge router will be assigned to such ESI. The ESI will be administratively configured on the DC GWs according to the procedures in [EVPN] and its use assumes that the DC GWs are connected to a single DC and to a single WAN domain. Multi-homing for cases where the DC GWs are connected to more than one DC and/or more than one WAN domain is for further study. This ESI will be referred as "DCI-ESI" hereafter. The solution (on the DC GWs) MUST follow the single-active multi- homing procedures as described in [EVPN-Overlays] for the provisioned DCI-ESI, i.e. Ethernet A-D routes per ESI and per EVI will be advertised to the DC NVEs. The MAC addresses learnt (in the data plane) on the hand-off links will be advertised with the DCI-ESI encoded in the ESI field. The use of OAM is recommended between the DC GWs and the WAN Edge routers: o If the DCI solution is based on a VLAN hand-off, 802.1ag/Y.1731 Ethernet-CFM can be used by the non-DF DC GW so that the peer WAN Edge router do not send any traffic to the DC GW for that particular EVI. o If the VPLS DCI solution is based on a pseudowire hand-off, the LDP PW Status bits TLV can be used by the non-DF to signal "Standby status" to the WAN Edge router for that particular EVI. Rabadan et al. Expires August 18, 2014 [Page 6] Internet-Draft DCI for E-VPN Overlays February 14, 2014 2.5. Data Center Gateway Optimizations The following features MAY be supported on the DC GW in order to optimize the control plane and data plane in the DC. 2.5.1 Use of the Unknown MAC route to reduce unknown flooding The use of EVPN, as the control plane of Network Virtualization Networks in the DC, brings a significant number of benefits as described in [EVPN-Overlays]. There are however some potential issues that SHOULD be addressed when the DC EVIs are connected to the WAN VPN instances. The first issue is the additional unknown unicast flooding created in the DC due to the unknown MACs existing beyond the DC GW. In virtualized DCs where all the MAC addresses are learnt in the control/management plane, unknown unicast flooding is significantly reduced. This is no longer true if the DC GW is connected to a layer-2 domain with data plane learning. The solution suggested in this document is based on the use of an "Unknown MAC route" that is advertised by the Designated Forwarder DC GW. The Unknown MAC route is a regular EVPN MAC/IP Advertisement route where the MAC Address Length is set to 48 and the MAC address to 00:00:00:00:00:00 (IP length is set to 0). If this procedure is used, when an EVI is created in the DC GWs and the Designated Forwarder (DF) is elected, the DF will send the Unknown MAC route. The NVEs supporting this concept will prune their unknown unicast flooding list and will only send the unknown unicast packets to the owner of the Unknown MAC route. Note that the DCI-ESI will be encoded in the ESI field of the NLRI so that regular multi-homing procedures can be applied to this unknown MAC too (e.g. backup-path). 2.5.2. MAC address advertisement control Another issue derived from the EVI interconnect to the WAN layer-2 domain is the potential massive MAC advertisement into the DC. All the MAC addresses learnt from the WAN on the hand-off attachment circuits or pseudowires must be advertised by BGP EVPN. Even if optimized BGP techniques like RT-constraint are used, the amount of MAC addresses to advertise or withdraw (in case of failure) from the DC GWs can be difficult to control and overwhelming for the DC network, especially when the NVEs reside in the hypervisors. This document proposes the addition of administrative options so Rabadan et al. Expires August 18, 2014 [Page 7] Internet-Draft DCI for E-VPN Overlays February 14, 2014 that the user can enable/disable the advertisement of MAC addresses learnt from the WAN as well as the advertisement of the Unknown MAC route from the DF DC GW. In cases where all the DC MAC addresses are learnt in the control/management plane, the DC GW may disable the advertisement of WAN MAC addresses. Any frame with unknown destination MAC will be exclusively sent to the Unknown MAC route owner(s). 2.5.3. ARP flooding control Another optimization mechanism, naturally provided by EVPN in the DC GWs, is the Proxy ARP function. The DC GWs SHOULD build a Proxy ARP cache table as per [EVPN]. When the active DC GW receives an ARP request coming from the WAN, the DC GW does a Proxy ARP table lookup and replies to the ARP request as long as the information is available in its table. This mechanism is specially recommended on the DC GWs since it protects the DC network from external ARP-flooding. 3. Integrated DCI solution for EVPN overlay networks When the DC and the WAN are operated by the same administrative entity, the Service Provider can decide to integrate the DC GW and WAN Edge PE functions in the same router for obvious CAPEX and OPEX saving reasons. This is illustrated in Figure 2. Note that this model does not provide an explicit demarcation link between DC and WAN anymore. ACLs or QoS policies between DC and WAN are not required. Rabadan et al. Expires August 18, 2014 [Page 8] Internet-Draft DCI for E-VPN Overlays February 14, 2014 +--+ |CE| +--+ | +----+ +----| PE |----+ +---------+ | +----+ | +---------+ +----+ | +---+ +---+ | +----+ |NVE1|--| |DC | |DC | |--|NVE3| +----+ | |GW1| |GW3| | +----+ | +---+ +---+ | | DC1 | WAN | DC2 | | +---+ +---+ | | |DC | |DC | | +----+ | |GW2| |GW4| | +----+ |NVE2|--| +---+ +---+ |--|NVE4| +----+ +---------+ | | +---------+ +----+ +--------------+ |<--EVPN-Overlay--->|<-----VPLS--->|<---EVPN-Overlay-->| |<--PBB-VPLS-->| DCI options -> |<-EVPN-MPLS-->| |<--PBB-EVPN-->| Figure 2 Integrated DCI model 3.1. Interconnect requirements The solution must observe the following requirements: o The DC GW function must provide control plane and data plane interworking between the EVPN-overlay network and the L2VPN technology supported in the WAN, i.e. (PBB-)VPLS or (PBB-)EVPN, as depicted in Figure 2. o Multi-homing MUST be supported. Single-active multi-homing with per-service load balancing MUST be implemented. All-active multi- homing, i.e. per-flow load-balancing, MUST be implemented as long as the technology deployed in the WAN supports it. o If EVPN is deployed in the WAN, the MAC Mobility, Static MAC protection and other procedures (e.g. proxy-arp) described in [EVPN] must be supported end-to-end. o Any type of inclusive multicast tree MUST be independently supported in the WAN as per [EVPN], and in the DC as per [EVPN- Overlays]. Rabadan et al. Expires August 18, 2014 [Page 9] Internet-Draft DCI for E-VPN Overlays February 14, 2014 3.2. VPLS DCI for EVPN-Overlay networks 3.2.1. Control/Data Plane setup procedures on the DC GWs Regular MPLS tunnels and TLDP/BGP sessions will be setup to the WAN PEs and RRs as per [RFC4761][RFC4762][RFC6074] and overlay tunnels and EVPN will be setup as per [EVPN-Overlays]. Note that different route-targets for the DC and for the WAN are normally required. A single type-1 RD per service can be used. In order to support multi-homing, the DC GWs will be provisioned with a DCI-ESI (see section 2.4), that will be unique per interconnection. Note that Ethernet Segment is a system wide assigned value, as opposed to the Ethernet Segments defined in [EVPN]. All the [EVPN] procedures are still followed for the DCI-ESI, e.g. any MAC address learnt from the WAN will be advertised to the DC with the DCI-ESI in the ESI field. A MAC-VRF per EVI will be created in each DC GW. The MAC-VRF will have two different types of tunnel bindings instantiated in two different split-horizon-groups: o VPLS pseudowires will be instantiated in the "WAN split-horizon-group". o Overlay tunnel bindings (e.g. VXLAN, NVGRE) will be instantiated in the "DC split-horizon-group". Attachment circuits are also supported on the same MAC-VRF, but they will not be part of any of the above split-horizon-groups. Traffic received in a given split-horizon-group will never be forwarded to a member of the same split-horizon-group. As far as BUM flooding is concerned, a flooding list will be created with the sub-list created by the inclusive multicast routes and the sub-list created for VPLS in the WAN. BUM frames received from a local attachment circuit will be flooded to both sub-lists. BUM frames received from the DC or the WAN will be forwarded to the flooding list observing the split-horizon-group rule described above. Note that the DC GWs are not allowed to have an EVPN binding and a pseudowire to the same far-end within the same MAC-VRF in order to avoid loops and packet duplication: o If an EVPN binding exists between two DC GWs and an attempt is made to setup a pseudowire between them, the pseudowire will be kept operationally down. The corresponding OAM signaling will be Rabadan et al. Expires August 18, 2014 [Page 10] Internet-Draft DCI for E-VPN Overlays February 14, 2014 triggered. o If a pseudowire exists between two DC GWs and an attempt is made to setup an EVPN binding, the pseudowire will be brought operationally down before establishing the EVPN binding. The optimizations procedures described in section 2.5 can also be applied to this option. 3.2.2. Multi-homing procedures on the DC GWs Single-active multi-homing MUST be supported on the DC GWs. All- active multi-homing is not supported by VPLS. All the single-active multi-homing procedures as described by [EVPN- Overlays] will be followed for the DCI-ESI. The non-DF DC GW for the DCI-ESI will block the transmission and reception of all the bindings in the "WAN aplit-horizon-group" for BUM and unicast traffic. 3.3. PBB-VPLS DCI for EVPN-Overlay networks 3.3.1. Control/Data Plane setup procedures on the DC GWs In this case, there is no impact on the procedures described in [RFC7041] for the B-component. However the I-component instances become EVI instances with EVPN-Overlay bindings and potentially local attachment circuits. M EVI instances can be multiplexed into the same B-component instance. This option provides significant savings in terms of pseudowires to be maintained in the WAN. The DCI-ESI concept described in section 3.2.1 will also be used for the PBB-VPLS-based DCI. B-component pseudowires and I-component EVPN-overlay bindings established to the same far-end will be compared. The following rules will be observed: o Attempts to setup a pseudowire between the two DC GWs within the B-component context will never be blocked. o If a pseudowire exists between two DC GWs for the B-component and an attempt is made to setup an EVPN binding on an I-component linked to that B-component, the EVPN binding will be kept operationally down. Note that the BGP EVPN routes will still be valid but not used. Rabadan et al. Expires August 18, 2014 [Page 11] Internet-Draft DCI for E-VPN Overlays February 14, 2014 o The EVPN binding will only be up and used as long as there is no pseudowire to the same far-end in the corresponding B-component. The EVPN bindings in the I-components will be brought down before the pseudowire in the B-component is brought up. The optimizations procedures described in section 2.5 can also be applied to this DCI option. 3.3.2. Multi-homing procedures on the DC GWs Single-active multi-homing MUST be supported on the DC GWs. All- active multi-homing MAY be supported. Procedures for the support of all-active multi-homing are for further study. All the single-active multi-homing procedures as described by [EVPN- Overlays] will be followed for the DCI-ESI for each EVI instance connected to B-component. The non-DF DC GW for the DCI-ESI will block the transmission and reception of all the EVPN bindings in the corresponding I-components for BUM and unicast traffic. 3.4. EVPN-MPLS DCI for EVPN-Overlay networks If EVPN for MPLS tunnels, EVPN-MPLS hereafter, is supported in the WAN, an end-to-end EVPN solution can be deployed. The following sections describe the proposed solution as well as the impact required on the [EVPN] procedures. 3.4.1. Control Plane setup procedures on the DC GWs The DC GWs MUST establish separate BGP sessions for sending/receiving EVPN routes to/from the DC and to/from the WAN. Normally each DC GW will setup one (two) BGP EVPN session(s) to the DC RR(s) and one(two) session(s) to the WAN RR(s). The route-distinguisher (RD) per EVI can be used for the EVPN routes sent to both, WAN and DC RRs. On the contrary, although reusing the same value is possible, different route-targets are expected to be handled for the same EVI in the WAN. As in the other discussed options, a DCI-ESI will be configured on the DC GWs for multi-homing. Received EVPN routes will never be reflected on the DC GWs but consumed and re-advertised (if needed): o Ethernet A-D routes, ES routes and inclusive multicast routes are consumed by the DC GWs and processed locally for the Rabadan et al. Expires August 18, 2014 [Page 12] Internet-Draft DCI for E-VPN Overlays February 14, 2014 corresponding [EVPN] procedures. o MAC/IP advertisement routes will be received, imported and if they become active in the MAC FIB, the information will be re- advertised as a new route: + The RD will be the DC GW's RD for the service. + The ESI will be set to the DCI-ESI. + The Ethernet-tag will be 0 or a new value. + The MAC length, MAC address, IP Length and IP address values will be kept from the previously received NLRI. + The MPLS label will be 0 or a local label. + The appropriate RTs and [RFC5512] BGP Encapsulation extended community will be used according to [EVPN-Overlays]. The DC GWs will also generate the following local EVPN routes that will be sent to the DC and WAN, with their corresponding RT and [RFC5512] BGP Encapsulation extended community values: o ES route for the DCI-ESI. o Ethernet A-D routes per ESI and EVI for the DCI-ESI. o Inclusive multicast routes with independent tunnel type value for the WAN and DC. E.g. a P2MP LSP may be used in the WAN whereas ingress replication is used in the DC. o MAC/IP advertisement routes for MAC addresses learnt in local attachment circuits. Note that these routes will not include the DCI-ESI, but ESI=0 or different from 0 for local Ethernet Segments (ES). Note that each DC GW will receive two copies of each of the above routes generated by the peer DC GW (one copy for the DC encapsulation and one copy for the WAN encapsulation). This is the expected behavior on the DC GW: o ES and A-D (per ESI) routes: regular BGP selection will be applied. o Inclusive multicast routes: if the Ethernet Tag ID matches on both routes, regular BGP selection applies and only one route will be active. It is recommended to influence the BGP selection Rabadan et al. Expires August 18, 2014 [Page 13] Internet-Draft DCI for E-VPN Overlays February 14, 2014 so that the DC route is preferred. If the Ethernet Tag ID does not match, then BGP will consider them two separate routes. In that case, the EVI service will select the DC route. o MAC/IP advertisement routes for local attachment circuits: as above, the DC GW will select only one. The decision will be made at BGP or service level, depending on the Ethernet Tags. The optimizations procedures described in section 2.5 can also be applied to this option. 3.4.2. Data Plane setup procedures on the DC GWs The procedure explained at the end of the previous section will make sure there are no loops or packet duplication between the DC GWs of the same DC since only one EVPN binding will be setup in the data plane between the two nodes. As for the rest of the EVPN tunnel bindings, two flooding lists will be setup by each DC GW for the same MAC-VRF: o EVPN-overlay flooding list (composed of bindings to the remote NVEs or multicast tunnel to the NVEs). o EVPN-mpls flooding list (composed of MP2P and or LSM tunnel to the remote PEs) Each flooding list will be part of a separate split-horizon group. Traffic generated from a local AC can be flooded to both split-horizon-groups. Traffic from a binding of a split-horizon-group can be flooded to the other split-horizon-group and local ACs, but never to a member of its own split-horizon-group. 3.4.3. Multi-homing procedures on the DC GWs Single-active as well as all-active multi-homing MUST be supported. All the multi-homing procedures as described by [EVPN] will be followed for the DF election for DCI-ESI, as well as the backup-path (single-active) and aliasing (all-active) procedures on the remote PEs/NVEs. The following changes are required at the DC GW with respect to the DCI-ESI: o Single-active multi-homing; assuming a WAN split-horizon-group, a DC split-horizon-group and local ACs on the DC GWs: + Forwarding behavior on the non-DF: the non-DF MUST NOT forward BUM or unicast traffic received from a given split-horizon- Rabadan et al. Expires August 18, 2014 [Page 14] Internet-Draft DCI for E-VPN Overlays February 14, 2014 group to a member of his own split-horizon group or to the other split-horizon-group. Only forwarding to local ACs is allowed (as long as they are not part of an ES for which the node is non-DF). + Forwarding behavior on the DF: the DF MUST NOT forward BUM or unicast traffic received from a given split-horizon-group to a member of his own split-horizon group or to the non-DF. Forwarding to the other split-horizon-group and local ACs is allowed (as long as they are not part of an ES for which the node is non-DF). o All-active multi-homing; assuming a WAN split-horizon-group, a DC split-horizon-group and local ACs on the DC GWs: + Forwarding behavior on the non-DF: the non-DF follows the same behavior as the non-DF in the single-active case but only for BUM traffic. Unicast traffic received from a split-horizon- group MUST NOT be forwarded to a member of its own split- horizon-group but can be forwarded normally to the other split-horizon-group and local ACs. If a known unicast packet is identified as a "flooded" packet, the procedures for BUM traffic MUST be followed. + Forwarding behavior on the DF: the DF follows the same behavior as the DF in the single-active case but only for BUM traffic. Unicast traffic received from a split-horizon-group MUST NOT be forwarded to a member of its own split-horizon- group but can be forwarded normally to the other split- horizon-group and local ACs. If a known unicast packet is identified as a "flooded" packet, the procedures for BUM traffic MUST be followed. o No ESI label is required to be signaled for DCI-ESI for its use by the non-DF in the data path. This is possible because the non-DF and the DF will never forward BUM traffic (coming from a split-horizon-group) to each other. 3.4.4. Impact on MAC Mobility procedures Since the MAC/IP Advertisement routes are not reflected in the DC GWs but rather consumed and re-advertised if active, the MAC Mobility procedures can be constrained to each domain (DC or WAN) and resolved within each domain. In other words, if a MAC moves within the DC, the DC GW MUST NOT re-advertise the route to the WAN with a change in the sequence number. Only when the MAC moves from the WAN domain to the DC domain, the DC GW will re-advertise the MAC with a higher sequence number in the MAC Mobility extended community. In respect to the MAC Rabadan et al. Expires August 18, 2014 [Page 15] Internet-Draft DCI for E-VPN Overlays February 14, 2014 Mobility procedures described in [EVPN] the MAC addresses learnt from the NVEs in the local DC or on the local ACs will be considered as local. The sequence numbers MUST NOT be propagated between domains. The sticky bit indication in the MAC Mobility extended community MUST be propagated between domains. 3.4.5. Data Center Gateway optimizations All the Data Center Gateway optimizations described in section 2.5 MAY be applied to the DC GWs when the DCI is based on EVPN-MPLS. In particular, the use of the Unknown MAC route, as described in section 2.5.1, reduces the unknown flooding in the DC but also solves some transient packet duplication issues in cases of all-active multi-homing. This is explained in the following paragraph. Consider the diagram in Figure 2 for EVPN-MPLS DCI and all-active multi-homing, and the following sequence: a) MAC Address M1 is advertised from NVE3 in EVI-1. b) DC GW3 and DC GW4 learn M1 for EVI-1 and re-advertise M1 to the WAN with DCI2-ESI in the ESI field. c) DC GW1 and DC GW2 learn M1 and install DC GW3/GW4 as next-hops following the EVPN aliasing procedures. d) Before NVE1 learns M1, a packet arrives to NVE1 with destination M1. The packet is subsequently flooded. e) Since both DC GW1 and DC GW2 know M1, they both forward the packet to the WAN (hence creating packet duplication), unless there is an indication in the data plane that the packet has been flooded by NVE1. If the DC GWs signal the same VNI/VSID for MAC/IP advertisement and inclusive multicast routes for EVI-1, such data plane indication does not exist. This undesired situation can be avoided by the use of the Unknown MAC route. If this route is used, the NVEs will prune their unknown unicast flooding list, and the non-DF DC GW will not received unknown packets, only the DF will. This solves the MAC duplication issue described above. 3.4.6. Benefits of the EVPN-MPLS DCI solution Besides retaining the EVPN attributes between Data Centers and Rabadan et al. Expires August 18, 2014 [Page 16] Internet-Draft DCI for E-VPN Overlays February 14, 2014 throughout the WAN, the EVPN-MPLS DCI solution on the DC GWs has some benefits compared to pure BGP EVPN RR or Inter-AS model B solutions without a gateway: o The solution supports the connectivity of local attachment circuits on the DC GWs. o Different data plane encapsulations can be supported in the DC and the WAN. o Optimized multicast solution, with independent inclusive multicast trees in DC and WAN. o MPLS Label aggregation: for the case where MPLS labels are signaled from the NVEs for MAC/IP Advertisement routes, this solution provides label aggregation. A remote PE MAY receive a single label per DC GW MAC-VRF as opposed to a label per NVE. o The DC GW will not propagate MAC mobility for the MACs moving within a DC. Mobility intra-DC is solved by all the NVEs in the DC. The MAC Mobility procedures on the DC GWs are only required in case of mobility across DCs. o Proxy-ARP function on the DGWs can be leveraged to reduce ARP flooding in the DC or/and in the WAN. 3.5. PBB-EVPN DCI for EVPN-Overlay networks [PBB-EVPN] is yet another DCI option. It requires the use of DC GWs where I-components and associated B-components are EVI instances. 3.5.1. Control/Data Plane setup procedures on the DC GWs EVPN will independently run in both components, the I-component EVI and B-component EVI. Compared to [PBB-EVPN], the DC C-MACs are no longer learnt in the data plane on the DC GW but in the control plane through EVPN running on the I-component. Remote C-MACs coming from remote PEs are still learnt in the data plane. B-MACs in the B- component will be assigned and advertised following the procedures described in [PBB-EVPN]. A DCI-ESI will be configured on the DC GWs for multi-homing, but it will only be used in the EVPN control plane for the I-component EVI. No ESI will be used in the control plane of the B-component EVI as per [PBB-EVPN]. The rest of the control plane procedures will follow [EVPN] for the I-component EVI and [PBB-EVPN] for the B-component EVI. Rabadan et al. Expires August 18, 2014 [Page 17] Internet-Draft DCI for E-VPN Overlays February 14, 2014 From the data plane perspective, the I-component and B-component EVPN bindings established to the same far-end will be compared and the I- component EVPN-overlay binding will be kept down following the rules described in section 3.3.1. 3.5.2. Multi-homing procedures on the DC GWs Single-active as well as all-active multi-homing MUST be supported. The forwarding behavior of the DF and non-DF will be changed based on the description outlined in section 3.4.3, only replacing the "WAN split-horizon-group" for the B-component. 3.5.3. Impact on MAC Mobility procedures C-MACs learnt from the B-component will be advertised in EVPN within the I-component EVI scope. If the C-MAC was previously known in the I-component database, EVPN would advertise the C-MAC with a higher sequence number, as per [EVPN]. From a Mobility perspective and the related procedures described in [EVPN], the C-MACs learnt from the B- component are considered local. 3.5.4. Data Center Gateway optimizations All the considerations explained in section 3.4.5 are applicable to the PBB-EVPN DCI option. 5. Conventions and Terminology 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 RFC-2119 [RFC2119]. BUM: it refers to the Broadcast, Unknown unicast and Multicast traffic DF: Designated Forwarder DC GW: Data Center Gateway DCI: Data Center Interconnect ES: Ethernet Segment ESI: Ethernet Segment Identifier DCI-ESI: ESI defined on the DC GWs for multi-homing to/from the WAN Rabadan et al. Expires August 18, 2014 [Page 18] Internet-Draft DCI for E-VPN Overlays February 14, 2014 EVI: EVPN Instance MAC-VRF: it refers to an EVI instance in a particular node NVE: Network Virtualization Edge TOR: Top-Of-Rack switch VNI/VSID: refers to VXLAN/NVGRE virtual identifiers 6. Security Considerations This section will be completed in future versions. 7. IANA Considerations 8. References 8.1. Normative References [RFC4761]Kompella, K., Ed., and Y. Rekhter, Ed., "Virtual Private LAN Service (VPLS) Using BGP for Auto-Discovery and Signaling", RFC 4761, January 2007. [RFC4762]Lasserre, M., Ed., and V. Kompella, Ed., "Virtual Private LAN Service (VPLS) Using Label Distribution Protocol (LDP) Signaling", RFC 4762, January 2007. [RFC6074]Rosen, E., Davie, B., Radoaca, V., and W. Luo, "Provisioning, Auto-Discovery, and Signaling in Layer 2 Virtual Private Networks (L2VPNs)", RFC 6074, January 2011. 8.2. Informative References [E-VPN] Sajassi et al., "BGP MPLS Based Ethernet VPN", draft-ietf- l2vpn-evpn-05.txt, work in progress, February, 2014 [PBB-EVPN] Sajassi et al., "PBB-EVPN", draft-ietf-l2vpn-pbb-evpn-06, work in progress, October, 2014 [EVPN-Overlays] Sajassi-Drake et al., "A Network Virtualization Overlay Solution using EVPN", draft-sd-l2vpn-evpn-overlay-02.txt, work in progress, October, 2013 Rabadan et al. Expires August 18, 2014 [Page 19] Internet-Draft DCI for E-VPN Overlays February 14, 2014 9. Acknowledgments This document was prepared using 2-Word-v2.0.template.dot. 10. Authors' Addresses Jorge Rabadan Alcatel-Lucent 777 E. Middlefield Road Mountain View, CA 94043 USA Email: jorge.rabadan@alcatel-lucent.com Senthil Sathappan Alcatel-Lucent Email: senthil.sathappan@alcatel-lucent.com Wim Henderickx Alcatel-Lucent Email: wim.henderickx@alcatel-lucent.com Florin Balus Nuage Networks Email: florin@nuagenetworks.net Senad Palislamovic Alcatel-Lucent Email: senad.palislamovic@alcatel-lucent.com Rabadan et al. Expires August 18, 2014 [Page 20]