Internet-Draft AC-DF per EVI July 2021
Wang Expires 12 January 2022 [Page]
Workgroup:
BESS WG
Published:
Intended Status:
Standards Track
Expires:
Author:
Y. Wang
ZTE Corporation

AC-Influenced DF Election per EVI

Abstract

The AC-influenced DF Election (AC-DF) per [RFC8584] is too dependent on EAD/EVI routes. For example, when no failures occured on an ESI, that AC-DF procedures will give incorrect results if no EAD/EVI routes are advertised. But in some light-weighted EVPNs (i.e. PBB EVPNs), no EAD/EVI routes will be advertised at all.

This draft proposes some new extensions to support AC-influenced DF Election without any EAD/EVI routes advertised in advance of any <ESI, EVI> failures.

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). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://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 12 January 2022.

1. Introduction

When the EAD/EVI route is not advertised before the corresponding ESI sub-interface fails, The AC-influenced DF Election procedures should elect the right DF before and after that failure.

Note that according to [RFC8584], the AC-influenced DF Election will be incorrect when no EAD/EVI route is advertised, even if no ESI sub-interface has failed at all.

This draft proposes some extension to DF-Capability negotiation and DF-Election procedures to support AC-influenced DF-election when no EAD/EVI routes are advertised.

1.1. Terminology

Most of the terminology used in this documents comes from [RFC8584] and [RFC7623] except for the following:

  • Light-weighted EVPN: The EVPN solution without EAD/EVI Route advertisedment.
  • EAD/ES: Ethernet A-D route per EVI, or RT-1 per ES route.
  • EAD/EVI: Ethernet A-D route per EVI, or RT-1 per EVI route.

2. AC-DF per EVI

2.1. Use Case

The ethernet segment ES1's ESI is ESI1, AC1/AC2 is two sub-interfaces on ES1, and AC3/AC4 is two sub-interfaces on ES2. AC1 and AC3 are attached to EVPN Instance EVI1, while AC2 and AC4 are attached to EVI2. The redundancy mode of ES1 is all-active.

                                 +----------+
                   PE1           |          |
              +-------------+    |          |
              | AC1         |    |          |         PE3
             /| AC2         |----|          |   +-------------+
            / |             |    |          |   |             |
       LAG /  +-------------+    |          |   |         AC3 |---CE2
   CE1=====                      |   PSN    |   |         AC4 |
           \  +-------------+    |          |---|             |
            \ |             |    |          |   +-------------+
             \| AC2         |----|          |
              | AC1         |    |          |
              +-------------+    |          |
                   PE2           |          |
                                 +----------+
Figure 1: AC-DF per EVI Usecase

EVI1 and EVI2 are two EVPN Instances, and there are no EAD/EVI routes advertised for them. Such EVPN Instances are called Light-weighted EVPNs in this draft. For example, EVI1 and EVI2 can be the I-Components of PBB EVPNs, because no EAD/EVI routes are advertised for these I-Components.

Initially PE1 is the DF for <ESI1, EVI1>, and PE2 is the DF for <ESI1, EVI2>. When the AC1 of PE1 fails (but ESI1 and AC2 of PE1 still works well), the DF for <ESI1, EVI1> should switch to PE2.

The DF-election should be done without any EAD/EVI routes advertised before any ESI sub-interface fails. Otherwise those EAD/EVI routes are advertised just for the adaptation of AC-influenced DF-election procedures per [RFC8584], but will do nothing good for the unicast packet forwarding in data plane (because data plane MAC learning don't use EAD/EVI routes).

2.2. AC-DF per EVI Capability

We introduce a new bit named "AC-DF per EVI" in the "bitmap" field of the DF Election extended community. The "AC-DF per EVI" bit means that the DF-Election for an <ESI, EVI> will not take EAD/EVI routes into considerations untill a Reverse EAD/EVI route for that <ESI, EVI> is received from any of the PEs.

2.2.1. Capability Negotiation Procedures

Only when all RT-4 routes of the same ESI indicate the "AC-DF per EVI" Capability, The DF Election will be executed in "AC-DF per EVI" mode. We can say that the DF Election mode for that ESI is negotiated as "AC-DF per EVI" mode.

Note that when any of the PEs on that ES is an old PE that don't support "AC-DF per EVI" mode, the RT-4 route from that PE will not indicate the "AC-DF per EVI" Capability, So the DF election will not be executed in "AD-DF per EVI" mode. This is for compatibility purpose.

2.2.2. AC-DF Capability vs AC-DF per EVI Capability

When all RT-4 routes of the same ESI indicate both the "AC-DF per EVI" Capability and the old AC-DF Capability, The DF Election will be executed in "AD-DF per EVI" mode.

Note that when any of the PEs on that ES is an old PE that don't support "AC-DF per EVI" mode, the RT-4 route from that PE may indicate the "old AC-DF" Capability only, So the DF election will still be executed in old AD-DF mode per [RFC8584]. This is for compatibility purpose.

2.3. DF Election Procedures

2.3.1. Initiation of AC-DF per EVI Mode

According to [RFC8584], the AC-influenced DF Election will be incorrect when no EAD/EVI route is advertised, even if no ESI sub-interface has failed at all.

But when the DF Election mode for an ESI is negotiated as AC-DF per EVI mode, no normal EAD/EVI routes can impact the DF Election procedures. The DF election will be done following that ESI's RT-4 routes only until at least one reverse EAD/EVI route is received.

2.3.2. Reverse EAD/EVI Route

In order to do AC-influenced DF election after a sub-interface of that ES fails, we introduce the "Reverse EAD/EVI Route". The reverse EAD/EVI Route is a special type of EAD/EVI Route that is advertised on the failure of corresponding <ESI, EVI>, not on the activation of that <ESI, EVI>.

Reverse EAD/EVI routes can use the same format as EAD/EVI Routes except for the following differences:

  • A Reverse EAD/EVI Route carries an EVPN Layer 2 Attributes Extended Community whose "Control Flags" field includes a new flag named "AC Down". The "AC Down" flag means that the corresponding AC (for which the Reverse EAD/EVI route is advertised) is down.

  • It is recommended to carry an ES-Import RT ([RFC7432]) and an EVI-RT ([I-D.ietf-bess-evpn-igmp-mld-proxy]) along with a reverse EAD/EVI route, no traditional Route-Targets have to be carried for DF-election purpose.

  • A Reverse EAD/EVI Route should make its MPLS label field be set to zero.

Note that when the corresponding sub-interface fails, the MP_REACH_NLRI of the reverse EAD/EVI route is advertised, and when the corresponding sub-interface recovers, the MP_UNREACH_NLRI of the reverse EAD/EVI route is sent. This is the opposite of normal EAD/EVI route. So it is called as reverse EAD/EVI route.

2.3.3. DF Election Rules

When a Reverse EAD/EVI Route for an <ESI, EVI> is received from a remote PE X, the RT-4 Route of that PE x are expelled from that <ESI, EVI>'s DF election. Then the DF election for that <ESI, EVI> will be updated according to the corresponding DF Alg.

Note that the DF election for other <ESI, EVI>s will not be affected by that Reverse EAD/EVI Route.

2.3.4. Route Filtering and RT Constraints Mechanism

When PE Y receives a reverse EAD/EVI route from PE X, and the ES-Import RT of that route can't match any local ES of PE Y, PE Y will not import that route into the EVI that is identified by that route's EVI-RT.

When RT Constraints Mechanism is enabled, each reverse EAD/EVI route will be distributed to the adjacent PEs of its ES only. Because that only the ES-Import RT are visible to the RT Constraints Mechanism, The EVI-RT is not visible to the RT Constraints Mechanism.

3. Other considerations

3.1. Reverse EAD/EVI Route in PBB EVPNs

The AC-DF per EVI mode is not confined to PBB EVPN which is just an example of light-weighted EVPNs. But in PBB EVPN, the construction of reverse EAD/EVI route need some special considerations.

  • It's EVI-RT should be the export route-target of the B-Component, not the C-Component.

  • It's Ethernet Tag ID (ETI) should be the I-SID of the I-Component.

Note that when PE Z receives a reverse EAD/EVI route whose EVI-RT matches a local B-Component but whose ETI matches none of the I-Components of that B-Component, PE Z may not import that reverse EAD/EVI route.

Note that the reverse EAD/EVI route don't have to carry any B-MAC along with it. Because that the B-MAC can do nothing helpful for the DF election.

3.2. Why no EAD/EVI Routes Advertised

When no RT-2 Routes advertised, no EAD/EVI routes need to be advertised either. PBB EVPN is an example of that. In PBB EVPN, the C-MACs are learnt in the data plane.

Other light-weighted EVPNs also do data plane C-MAC learning, so they don't have to advertise EAD/EVI routes either. In such EVPNs, AC-DF per EVI will help.

4. Comparison with other solutions

PBB EVPN can assign a deicated vES to each sub-interface, in such case, the RT-4 routes are advertised per each sub-interface (or vESI). But this will bring out some other disadvantages:

  • The uniformity of service carving can't be achieved without careful configuring.

    With service carving, it is possible to elect multiple DFs per ES (one per EVI) in order to perform load balancing of traffic destined to a given ES. The objective is that the load-balancing procedures should carve up the BD space among the redundant PE nodes evenly, in such a way that every PE is the DF for a distinct set of EVIs.

    When each EVI use a dedicated vESI to advertise the corresponding Ethernet Segment Routes for that <ES, EVI>, The service carving mechanisms can not work without manual configuration.

  • The amount of B-MACs will be greatly increased.

The brief comparisons are listed as the following table:

Table 1: Comparisons with vESI for []
Items AC-DF per EVI vESI
ESIs one per port one per sub-interface
RT-4 Routes one per port one per sub-interface
B-MACs one per port one per sub-interface
Service Carving auto manual-configured
EAD per EVI routes none none
Reverse EAD per EVI routes one per failed sub-interfaces none

Using AC-DF per EVI mode, the service-carving is automatically achieved, and no extra B-MACs should be configured and advertised.

5. Security Considerations

Security considerations will be added in future versions.

6. IANA Considerations

6.1. New DF Election Capability

IANA will be requested to allocate a new DF Election Capability in the "DF Election Capabilities" Registry. This capability is called "AC-DF per EVI Capability".


      Bit         Name                             Reference
      ----        ----------------                 -------------
      4           AC-DF per EVI Capability         This draft

Figure 2: AC-DF per EVI Capability

6.2. New EVPN Layer 2 Attributes Control Flags

IANA will be requested to allocate a new Control Flag in the "EVPN Layer 2 Attributes Control Flags" Registry. This Control Flag is called "D" Flag, where "D" means AC-Down.


   Bit     Name                                        Reference
   ----    ----------------                            -------------
   D       AC-Down on Advertising PE                   This Draft

Figure 3: AC Failure Flag

7. Acknowledgements

The authors would like to thank the following for their comments and review of this document:

Chunning Dai.

8. Normative References

[I-D.ietf-bess-evpn-igmp-mld-proxy]
Sajassi, A., Thoria, S., Mishra, M. P., Patel, K., Drake, J., and W. Lin, "IGMP and MLD Proxy for EVPN", Work in Progress, Internet-Draft, draft-ietf-bess-evpn-igmp-mld-proxy-09, , <https://datatracker.ietf.org/doc/html/draft-ietf-bess-evpn-igmp-mld-proxy-09>.
[I-D.ietf-bess-srv6-services]
Dawra, G., Filsfils, C., Talaulikar, K., Raszuk, R., Decraene, B., Zhuang, S., and J. Rabadan, "SRv6 BGP based Overlay Services", Work in Progress, Internet-Draft, draft-ietf-bess-srv6-services-07, , <https://datatracker.ietf.org/doc/html/draft-ietf-bess-srv6-services-07>.
[RFC7432]
Sajassi, A., Ed., Aggarwal, R., Bitar, N., Isaac, A., Uttaro, J., Drake, J., and W. Henderickx, "BGP MPLS-Based Ethernet VPN", RFC 7432, DOI 10.17487/RFC7432, , <https://www.rfc-editor.org/info/rfc7432>.
[RFC7623]
Sajassi, A., Ed., Salam, S., Bitar, N., Isaac, A., and W. Henderickx, "Provider Backbone Bridging Combined with Ethernet VPN (PBB-EVPN)", RFC 7623, DOI 10.17487/RFC7623, , <https://www.rfc-editor.org/info/rfc7623>.
[RFC8584]
Rabadan, J., Ed., Mohanty, S., Ed., Sajassi, A., Drake, J., Nagaraj, K., and S. Sathappan, "Framework for Ethernet VPN Designated Forwarder Election Extensibility", RFC 8584, DOI 10.17487/RFC8584, , <https://www.rfc-editor.org/info/rfc8584>.

9. Informative References

[RFC7041]
Balus, F., Ed., Sajassi, A., Ed., and N. Bitar, Ed., "Extensions to the Virtual Private LAN Service (VPLS) Provider Edge (PE) Model for Provider Backbone Bridging", RFC 7041, DOI 10.17487/RFC7041, , <https://www.rfc-editor.org/info/rfc7041>.
[RFC7348]
Mahalingam, M., Dutt, D., Duda, K., Agarwal, P., Kreeger, L., Sridhar, T., Bursell, M., and C. Wright, "Virtual eXtensible Local Area Network (VXLAN): A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks", RFC 7348, DOI 10.17487/RFC7348, , <https://www.rfc-editor.org/info/rfc7348>.
[RFC8365]
Sajassi, A., Ed., Drake, J., Ed., Bitar, N., Shekhar, R., Uttaro, J., and W. Henderickx, "A Network Virtualization Overlay Solution Using Ethernet VPN (EVPN)", RFC 8365, DOI 10.17487/RFC8365, , <https://www.rfc-editor.org/info/rfc8365>.

Author's Address

Yubao Wang
ZTE Corporation
No.68 of Zijinghua Road, Yuhuatai Distinct
Nanjing
China