RFC 8380

Internet Engineering Task Force (IETF)                         L. Dunbar
Request for Comments: 8380                               D. Eastlake 3rd
Category: Standards Track                                         Huawei
ISSN: 2070-1721                                               R. Perlman
                                                                May 2018

   Transparent Interconnection of Lots of Links (TRILL) Encapsulation


   This document describes how data center networks can benefit from
   non-RBridge nodes performing TRILL (Transparent Interconnection of
   Lots of Links) encapsulation with assistance from a directory

Status of This Memo

   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at

Copyright Notice

   Copyright (c) 2018 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
   (https://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.

Dunbar, et al.               Standards Track                    [Page 1]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

Table of Contents

   1. Introduction ....................................................2
   2. Conventions Used in This Document ...............................2
   3. Directory Assistance to Non-RBridge .............................3
   4. Source Nickname in Encapsulation by Non-RBridge Nodes ...........6
   5. Benefits of a Non-RBridge Performing TRILL Encapsulation ........6
      5.1. Avoid Nickname Exhaustion Issue ............................6
      5.2. Reduce MAC Tables for Switches on Bridged LANs .............6
   6. Manageability Considerations ....................................7
   7. Security Considerations .........................................7
   8. IANA Considerations .............................................9
   9. References  .....................................................9
      9.1.  Normative References .....................................10
      9.2.  Informative References ...................................10
   Acknowledgments ...................................................10
   Authors' Addresses.................................................10

1.  Introduction

   This document describes how data center networks can benefit from
   non-RBridge nodes performing TRILL encapsulation with assistance from
   a directory service and specifies a method for them to do so.

   [RFC7067] and [RFC8171] describe the framework and methods for edge
   RBridges to get (MAC and VLAN) <-> Edge RBridge mapping from a
   directory service instead of flooding unknown destination MAC
   addresses across a TRILL domain.  If it has the needed directory
   information, any node, even a non-RBridge node, can perform the TRILL
   data packet encapsulation.  This document describes the benefits of
   and a scheme for non-RBridge nodes performing TRILL encapsulation.

2.  Conventions Used in This Document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "OPTIONAL" in this document are to be interpreted as described in
   BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

   AF:       Appointed Forwarder RBridge port [RFC8139].

   Bridge:   A device compliant with IEEE 802.1Q.  In this document,
             Bridge is used interchangeably with Layer 2 switch.

   DA:       Destination Address.

   ES-IS:    End System to Intermediate System [RFC8171].

Dunbar, et al.               Standards Track                    [Page 2]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

   Host:     A physical server or a virtual machine running
             applications.  A host usually has at least one IP address
             and at least one MAC address.

   IS-IS:    Intermediate System to Intermediate System [RFC7176].

   SA:       Source Address.

   TRILL-EN: TRILL Encapsulating Node.  A node that performs the TRILL
             encapsulation but doesn't participate in an RBridge's IS-IS

   VM:       Virtual Machine.

3.  Directory Assistance to Non-RBridge

   With directory assistance [RFC7067] [RFC8171], a non-RBridge node can
   learn if a data packet needs to be forwarded across the RBridge
   domain and, if so, the corresponding egress RBridge.

   Suppose the RBridge domain boundary starts at network switches (not
   virtual switches embedded on servers).  (See Figure 1 for a high-
   level diagram of a typical data center network.)  A directory can
   assist virtual switches embedded on servers to encapsulate with a
   proper TRILL header by providing the nickname of the egress RBridge
   edge to which the destination is attached.  The other information
   needed to encapsulate can be learned either by listening to TRILL
   ES-IS and/or IS-IS Hellos [RFC7176] [RFC8171], which will indicate
   the MAC address and nickname of appropriate local edge RBridges, or
   by configuration.

   If it is not known whether a destination is attached to one or more
   edge RBridges, based on the directory, the non-RBridge node can
   forward the data frames natively, i.e., not encapsulating with any
   TRILL header.  Or, if the directory is known to be complete, the non-
   RBridge node can discard such data frames.

Dunbar, et al.               Standards Track                    [Page 3]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

          \           +-----------+       +-----------+            /
           \        +/----------+ |     +/----------+ |  TRILL    /
            \       |Aggregation| |     |Aggregation| | Domain   /
             \      |     11    | + --- |     N1    | +         /
              \     +-----------+/      +-----------+/         /
               \         /     \            /      \          /
                \       /       \          /        \        /
         Top-    \   +---+    +---+      +---+     +---+    /
         of- -->  \- |T11|... |T1x|      |T21| ..  |T2y|---/
         Rack        +---+    +---+      +---+     +---+
         Switches      |        |          |         |
                     +-|-+    +-|-+      +-|-+     +-|-+
                     |   |... | V |      | V | ..  | V | <- vSwitch
                     +---+    +---+      +---+     +---+
                     |   |... | V |      | V | ..  | V |
                     +---+    +---+      +---+     +---+
                     |   |... | V |      | V | ..  | V |
                     +---+    +---+      +---+     +---+

          Figure 1: TRILL Domain in a Typical Data Center Network

   When a TRILL-encapsulated data packet reaches the ingress RBridge,
   that RBridge simply performs the usual TRILL processing and forwards
   the pre-encapsulated packet to the RBridge that is specified by the
   egress nickname field of the TRILL header.  When an ingress RBridge
   receives a native Ethernet frame in an environment with complete
   directory information, the ingress RBridge doesn't flood or forward
   the received data frames when the destination MAC address in the
   Ethernet data frames is unknown.

   When all end nodes attached to an ingress RBridge pre-encapsulate
   with a TRILL header for traffic across the TRILL domain, the ingress
   RBridge doesn't need to encapsulate any native Ethernet frames to the
   TRILL domain.  The attached nodes can be connected to multiple edge
   RBridges by having multiple ports or through a bridged LAN.  All
   RBridge edge ports connected to one bridged LAN can receive and
   forward pre-encapsulated traffic; this can greatly improve the
   overall network utilization.  However, it is still necessary to, for
   example, designate AF ports to be sure that multi-destination packets
   from the TRILL campus are only egressed through one RBridge.

   Item 8 of Section 4.6.2 of the TRILL base protocol specification
   [RFC6325] specifies that an RBridge port can be configured to accept
   TRILL-encapsulated frames from a neighbor that is not an RBridge.

   When a TRILL frame arrives at an RBridge whose nickname matches the
   destination nickname in the TRILL header of the frame, the processing
   is exactly as normal: as specified in [RFC6325], the RBridge

Dunbar, et al.               Standards Track                    [Page 4]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

   decapsulates the received TRILL frame and forwards the decapsulated
   frame to the target attached to its edge ports.  When the destination
   MAC address of the decapsulated Ethernet frame is not in the egress
   RBridge's local MAC attachment tables, the egress RBridge floods the
   decapsulated frame to all attached links in the frame's VLAN, or
   drops the frame (if the egress RBridge is configured with that

   We call a node that, as specified herein, only performs TRILL
   encapsulation, but doesn't participate in RBridge's IS-IS routing, a
   TRILL Encapsulating Node (TRILL-EN).  The TRILL Encapsulating Node
   can pull (MAC and VLAN) <-> Edge RBridge mapping from directory
   servers [RFC8171].  In order to do this, a TRILL-EN MUST support
   TRILL ES-IS [RFC8171].

   Upon receiving or locally generating a native Ethernet frame, the
   TRILL-EN checks the (MAC and VLAN) <-> Edge RBridge mapping and
   performs the corresponding TRILL encapsulation if the mapping entry
   is found as shown in Figure 2.  If the destination MAC address and
   VLAN of the received Ethernet frame doesn't exist in the mapping
   table and there is no positive reply from pull requests to a
   directory, the Ethernet frame is dropped or is forwarded in native
   form to an edge RBridge, depending on the TRILL-EN configuration.

       |OuterEtherHd|TRILL HD| InnerDA | InnerSA |..|Payload|FCS|
               |             |<Inner Ether Header>  |
               |       +-------+  TRILL    +------+
               |       |  RB1  |---------->|  RB2 |  Decapsulate
               |       +-------+  domain   +------+  TRILL header
               v           ^                   |
               +---------->|                   |
                           |                   V
                        +--------+         +--------+
      Non-RBridge node: |TRILL-EN|         |TRILL-EN|
      Encapsulate TRILL |    1   |         |    2   |
      Header for data   +--------+         +--------+
      Frames to traverse TRILL domain.

                    Figure 2: Data Frames from a TRILL-EN

Dunbar, et al.               Standards Track                    [Page 5]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

4.  Source Nickname in Encapsulation by Non-RBridge Nodes

   The TRILL header includes a Source RBridge's Nickname (ingress) and
   Destination RBridge's Nickname (egress).  When a TRILL header is
   added to a data packet by a TRILL-EN, the ingress RBridge nickname
   field in the TRILL header is set to a nickname of the AF for the data
   packet's VLAN.  The TRILL-EN determines the AF by snooping on IS-IS
   Hellos from the edge RBridges on the link with the TRILL-EN in the
   same way that the RBridges on the link determine the AF [RFC8139].  A
   TRILL-EN is free to send the encapsulated data frame to any of the
   edge RBridges on its link.

5.  Benefits of a Non-RBridge Performing TRILL Encapsulation

   This section summarizes the benefits of having a non-RBridge node
   perform TRILL encapsulation.

5.1.  Avoid Nickname Exhaustion Issue

   For a large data center with hundreds of thousands of virtualized
   servers, setting the TRILL boundary at the servers' virtual switches
   will create a TRILL domain with hundreds of thousands of RBridge
   nodes; this could lead to TRILL nickname exhaustion and challenges to
   IS-IS.  On the other hand, setting the TRILL boundary at aggregation
   switches that have many virtualized servers attached can limit the
   number of RBridge nodes in a TRILL domain, but introduces the issue
   of having very large (MAC and VLAN) <-> Edge RBridge mapping tables
   that need to be maintained by edge RBridges.

   Allowing non-RBridge nodes to pre-encapsulate data frames with TRILL
   headers makes it possible to have a TRILL domain with a reasonable
   number of RBridge nodes in a large data center.  All the TRILL-ENs
   attached to one RBridge can be represented by one TRILL nickname,
   which can avoid the nickname exhaustion problem.

5.2.  Reduce MAC Tables for Switches on Bridged LANs

   When hosts in a VLAN (or subnet) span across multiple edge RBridges
   and each edge RBridge has multiple VLANs enabled, the switches on the
   bridged LANs attached to the edge RBridges are exposed to all MAC
   addresses among all the VLANs enabled.

   For example, for an Access Switch with 40 physical servers attached,
   where each server has 100 VMs, there are 4000 hosts under the Access
   Switch.  If indeed hosts/VMs can be moved anywhere, the worst case
   for the Access Switch is when all those 4000 VMs belong to different
   VLANs, i.e., the Access Switch has 4000 VLANs enabled.  If each VLAN

Dunbar, et al.               Standards Track                    [Page 6]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

   has 200 hosts, this Access Switch's MAC table potentially has
   200 * 4000 = 800,000 entries.

   If the virtual switches on servers pre-encapsulate the data frames
   destined for hosts attached to remote edge RBridges, the outer MAC
   destination address of those TRILL-encapsulated data frames will be
   the MAC address of a local RBridge edge, i.e., the ingress RBridge.
   The switches on the local bridged LAN don't need to keep the MAC
   entries for remote hosts attached to other edge RBridges.

   But the TRILL traffic from nodes attached to other RBridges is
   decapsulated and has the true source and destination MACs.  One
   simple way to prevent local bridges from learning remote hosts' MACs
   and adding to their MAC tables, if that would be a problem, is to
   disable this data-plane learning on local bridges.  With the
   assistance of a directory, the local bridges can be pre-configured
   with MAC addresses of local hosts.  The local bridges can always send
   frames with unknown destination MAC addresses to the ingress RBridge.
   In an environment where a large number of VMs are instantiated in one
   server, the number of remote MAC addresses could be very large.  If
   it is not feasible to disable learning and pre-configure MAC tables
   for local bridges and all important traffic is IP, one effective
   method to minimize local bridges' MAC table size is to use the
   server's MAC address to hide MAC addresses of the attached VMs.  That
   is, the server acting as an edge node uses its own MAC address in the
   source MAC address field of the packets originated from a host (or
   VM).  When the Ethernet frame arrives at the target edge node (the
   egress), the target edge node can send the packet to the
   corresponding destination host based on the packet's IP address.
   Very often, the target edge node communicates with the embedded VMs
   via a Layer 2 virtual switch.  In this case, the target edge node can
   construct the proper Ethernet header with the assistance of the
   directory.  The information from the directory includes the proper
   mapping of host IP to MAC.

6.  Manageability Considerations

   Directory assistance [RFC8171] is required to make it possible for a
   non-TRILL node to pre-encapsulate packets destined towards remote
   RBridges.  TRILL-ENs have the same configuration options as any pull
   directory client.  See Section 4 of [RFC8171].

7.  Security Considerations

   If the TRILL-ENs are not trusted, they can forge arbitrary ingress
   and egress nicknames in the TRILL Headers of the TRILL Data packets
   they construct.  With data-plane learning, decapsulating a TRILL Data
   packet at an egress RBridge associates the inner source MAC address

Dunbar, et al.               Standards Track                    [Page 7]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

   with the ingress nickname in the TRILL Header (assuming that MAC
   address is unicast).  Thus, if those ingress nicknames are forged,
   incorrect learning will occur and future traffic destined for the
   inner source MAC will be sent to the wrong RBridge for egress.
   Because of this, an RBridge port should not be configured to accept
   encapsulated TRILL data frames on a link were it does not have an
   RBridge adjacency unless the end stations on that link are trusted.

   As with any end station, TRILL-ENs can forge the outer MAC addresses
   of packets they send.  (See Section 6 of [RFC6325].) Because they
   pre-encapsulate, they can also forge inner MAC addresses.

   The pre-encapsulation performed by TRILL-ENs also means they can send
   data in any VLAN; this means they must be trusted in order to enforce
   a security policy based on VLANs.  (See Section 6.1 of [RFC6325].)

   Use of directory-assisted encapsulation by TRILL-ENs essentially
   involves those TRILL-ENs spoofing edge RBridges to which they are
   connected; this is another reason that TRILL-ENs should be trusted
   nodes.  Such spoofing cannot cause persistently looping traffic
   because TRILL has a hop count in the TRILL header [RFC6325] so that,
   should there be a loop, a TRILL packet caught in that loop (i.e., an
   encapsulated frame) will be discarded.  (In the potentially more
   dangerous case of multidestination packets (as compared with known
   unicast) where copies could multiply due to forks in the distribution
   tree, a Reverse Path Forwarding Check is also used [RFC6325] to
   discard packets that appear to be on the wrong link or when there is
   disagreement about the distribution tree.)

   The mechanism described in this document requires a TRILL-EN to be
   aware of the MAC address(es) of the TRILL edge RBridge(s) to which
   the TRILL-EN is attached and the egress RBridge nickname from which
   the destination of the packets is reachable.  With that information,
   TRILL-ENs can learn a substantial amount about the topology of the
   TRILL domain.  Therefore, there could be a potential security risk
   when the TRILL-ENs are not trusted or are compromised.

   If the path between the directory and a TRILL-EN is attacked, false
   mappings can be sent to the TRILL-EN causing packets from the TRILL-
   EN to be sent to wrong destinations, possibly violating security
   policy as to which end stations should receive what data.  Therefore,
   a combination of authentication and encryption is RECOMMENDED between
   the directory and TRILL-EN.  The entities involved will need to
   properly authenticate with each other, provide session encryption,
   maintain security patch levels, and configure their systems to allow
   minimal access and running processes to protect sensitive

Dunbar, et al.               Standards Track                    [Page 8]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

   For added security against the compromise of data due to its
   misdelivery for any reason, including the above, end-to-end
   encryption and authentication should be considered; that is,
   encryption and authentication from source end station to destination
   end station.

   For Pull Directory and TRILL ES-IS security considerations, see

8.  IANA Considerations

   This document has no IANA actions.

9.  References

9.1.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,

   [RFC6325]  Perlman, R., Eastlake 3rd, D., Dutt, D., Gai, S., and A.
              Ghanwani, "Routing Bridges (RBridges): Base Protocol
              Specification", RFC 6325, DOI 10.17487/RFC6325, July 2011,

   [RFC7176]  Eastlake 3rd, D., Senevirathne, T., Ghanwani, A., Dutt,
              D., and A. Banerjee, "Transparent Interconnection of Lots
              of Links (TRILL) Use of IS-IS", RFC 7176,
              DOI 10.17487/RFC7176, May 2014,

   [RFC8139]  Eastlake 3rd, D., Li, Y., Umair, M., Banerjee, A., and F.
              Hu, "Transparent Interconnection of Lots of Links (TRILL):
              Appointed Forwarders", RFC 8139, DOI 10.17487/RFC8139,
              June 2017, <https://www.rfc-editor.org/info/rfc8139>.

   [RFC8171]  Eastlake 3rd, D., Dunbar, L., Perlman, R., and Y. Li,
              "Transparent Interconnection of Lots of Links (TRILL):
              Edge Directory Assistance Mechanisms", RFC 8171,
              DOI 10.17487/RFC8171, June 2017,

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/info/rfc8174>.

Dunbar, et al.               Standards Track                    [Page 9]

RFC 8380             Directory-Assisted TRILL Encap             May 2018

9.2.  Informative References

   [RFC7067]  Dunbar, L., Eastlake 3rd, D., Perlman, R., and I.
              Gashinsky, "Directory Assistance Problem and High-Level
              Design Proposal", RFC 7067, DOI 10.17487/RFC7067, November
              2013, <https://www.rfc-editor.org/info/rfc7067>.


   The following are thanked for their contributions:

      Igor Gashinsky
      Ben Nevin-Jenkins

Authors' Addresses

   Linda Dunbar
   Huawei Technologies
   5340 Legacy Drive, Suite 175
   Plano, TX  75024
   United States of America

   Phone: +1-469-277-5840
   Email: linda.dunbar@huawei.com

   Donald Eastlake 3rd
   Huawei Technologies
   155 Beaver Street
   Milford, MA  01757
   United States of America

   Phone: +1-508-333-2270
   Email: d3e3e3@gmail.com

   Radia Perlman
   2010 256th Avenue NE, #200
   Bellevue, WA  98007
   United States of America

   Email: Radia@alum.mit.edu

Dunbar, et al.               Standards Track                   [Page 10]