| Internet-Draft | MNAT | October 2020 |
| Holland | Expires 4 May 2021 | [Page] |
This document defines a method for a network to maintain Network Address Translation address mappings for the transport of globally addressed multicast traffic within a network that can't otherwise forward the globally addressed traffic. A new Multicast Network Address Translation (MNAT) service is defined to communicate the address mappings to ingress and egress points within the network, and considerations for operation of the MNAT service are described.¶
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 4 May 2021.¶
Copyright (c) 2020 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.¶
Network Address Translation is very widely used for unicast traffic in a variety of networks and according to a variety of mechanisms. [RFC2663] is recommended reading for background on the ways unicast NAT is used.¶
The handling of multicast traffic can pose a variety of additional problems for a network, some of which can be mitigated or avoided if traffic can be mapped to a different address space than its original addressing. This document defines a new service, Multicast Network Address Translation (MNAT), as a mechanism to administer network address mappings for multicast traffic within a network, for the purpose of working around various addressing-related issues. An overview of some of the motivating use cases that require network address remapping for multicast traffic is given in Section 1.3. An explanation of the protocol operation is given in Section 2.¶
Messaging to and from the MNAT service is defined with RESTCONF [RFC8040] using the YANG [RFC7950] model in Section 3.¶
Unlike traditional unicast NAT, MNAT performs address translation at both an ingress point to the network where the traffic is transformed to use an address scheme local to the network, and also at an egress point from the network where the traffic is transformed back to the original address scheme for further forwarding, or for further processing by a receiving application.¶
The reader is assumed to be familiar with the concepts and terminology regarding source-specific multicast as described in [RFC4607] and the use of IGMPv3 [RFC3376] and MLDv2 [RFC3810] for group management of source-specific multicast channels, as described in [RFC4604].¶
The reader is also assumed to be familiar with the concepts and terminology for RESTCONF [RFC8040] and YANG [RFC7950].¶
The reader is also assumed to be familiar with the use of DNS-SD [RFC6763] for discovery of services provided by the network to end hosts.¶
| Term | Definition |
|---|---|
| (S,G) | A source-specific multicast channel, as described in [RFC4607]. A pair of IP addresses with a source host IP and destination group IP. |
| egress node | A MNAT client operating at a point where NATted multicast traffic exits the network. |
| ingress node | A MNAT client operating at a point where multicast traffic enters the network and gets NATted |
| MNAT client | A client using the ietf-mnat YANG model via RESTCONF, or a client with equivalent signaling to an MNAT service. |
| NATted traffic | Multicast traffic that has been translated to use addressing or encapsulation assigned locally within the network, rather than its original global addressing. |
| SSM | Source-specific multicast, as described in [RFC4607] |
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119] and [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This section lists use cases where a global (S,G) may not be possible to transport within a network, requiring the use of some kind of encapsulation or address translation in order to adequately communicate the group membership for packet replication within the network, or in order to perform the forwarding for the subscribed traffic within the network.¶
A note elaborating on the use of static provisioning of multicast groups:¶
Some networks have found that there are good use cases to deliver a limited set of packet-replicating flows, including sometimes the use of externally sourced multicast traffic, but have struggled with the operational complexity of operating a dynamic tree-building system based on PIM-SM [RFC7761]. Operating an MNAT service can allow these networks to provide for the limited use of packet-replicating data channels while keeping the operational complexity of handling a dynamically changing set of channels confined to a single service that implements their business logic for admission control, rather than trying to apply access control lists for group membership propagation spread across the network.¶
Note to RFC Editor: Please remove this section and its subsections before publication.¶
This section is to provide references to make it easier to review the development and discussion on the draft so far.¶
This document is in the Github repository at:¶
https://github.com/GrumpyOldTroll/draft-ietf-mnat¶
Readers are welcome to open issues and send pull requests for this document.¶
Please note that contributions may be merged and substantially edited, and as a reminder, please carefully consider the Note Well before contributing: https://datatracker.ietf.org/submit/note-well/¶
Substantial discussion of this document should take place on the MBONED working group mailing list (mboned@ietf.org).¶
The use of MNAT within a network is defined in terms the folowing entities:¶
Address translation is performed at the ingress and egress nodes. Ingress is where an external (S,G) is mapped to locally assigned address mapping before being forwarded for transport within the network. Egress is where the traffic received on locally assigned addresses is translated back to the corresponding external (S,G) address before being forwarded for further transmission or processed by a receiving application.¶
The MNAT service maintains the mapping between external (S,G)s and the local network addresses used to transport traffic of those (S,G)s within the network. The address mapping is performed according to the needs of the network operating the MNAT service, to satisfy whatever constraints and restrictions may be necessary or desirable according to the operational considerations within that network. Some example considerations that have motivated the design of MNAT are described in Section 1.3.¶
Ingress and egress nodes communicate with the MNAT service according to the schema defined by the YANG model in Section 3. In particular, they maintain an up-to-date table of the mappings between the external (S,G)s and the locally assigned addresses for transport within the network in order to perform the corresponding network address translations.¶
TBD: probably add a diagram here. Probably something roughly similar to page 7 of the IETF 108 mboned presentation touching on this: https://www.ietf.org/proceedings/108/slides/slides-108-mboned-status-update-on-multicast-to-the-browser-00.pdf#page=7¶
Egress nodes can run in at least two separate modes of operation.¶
One of the modes is "bump in the wire", which refers to a node that receives traffic using the network-assigned locally chosen addresses, and translates the traffic back to the associated externally addressed (S,G) before forwarding the traffic along the rest of the network paths to the receiving applications that tried to join the external (S,G).¶
The second mode is "bump in the host", which refers to a virtual node operating inside a client application.¶
As a "bump in the host" egress node, the virtual egress node can discover and connect to the MNAT service from a receiving application. The receiving application would then use the knowledge about the address mapping within the network to perform a join for the mapped addresses in the local network, rather than for the external (S,G), treating the payloads of the packets received as though they arrived with the external (S,G) addressing.¶
A common scenario for a bump in the wire egress node deployment might be to have egress nodes operating in Customer Premesis Equipment (CPE), such as a Cable Modem or Wi-Fi router inside the home of a customer to a multicast-capable Internet Service Provider (ISP). In this scenario, the egress node discovery mechanism for the MNAT service might be a static configuration for the MNAT service's hostname, pushed by the ISP to the CPE devices.¶
For a bump in the host egress node, the discovery of the MNAT service might either operate via DNS-SD [RFC6763] using a search domain for the ISP distributed to hosts via a DHCP Domain Search option [RFC3397], or via configuration instructions the ISP gives to their customers to configure a search domain for their devices, or to configure the MNAT service's hostname for that ISP in their applications.¶
It is RECOMMENDED that a network operating an MNAT service provide service discovery with the use of DNS-SD [RFC6763]. However, a network MAY use other mechanisms, including options such as manual configuration. As long as an MNAT client can find a valid hostname to use, it can connect to the given MNAT service and monitor changes to the address assignments within the network.¶
TBD: recommendations for noticing and discontinuing use of MNAT services that report mappings that don't correspond to the mappings apparently in use in the client's local network (particularly from egress nodes).¶
TBD: describe the RESTCONF validation and bootstrapping steps. Use the same section name from I-D.draft-ietf-mboned-dorms as a template, assuming it passes a wider review.¶
When possible, changes to the group assignments should be communicated with subscriptions to data model updates using a server push mechanism, for example as described in [RFC8641].¶
Where clients or servers do not support server push updates, long polling can be used instead to provide timely updates. See [RFC6202] for an explanation of the approach and a discussion of its pros and cons.¶
If long polling and server push are both unavailable, MNAT clients may need to poll the server to monitor updates instead. This approach is likely to encounter delays in the detection of changes to mapping decisions within the MNAT service, but can be used as a last resort for providing multicast connectivity.¶
Egress nodes open a persistent connection to the MNAT service and request allocation of an egress key with the get-new-egress-key rpc. Egress keys are identifiers chosen by the MNAT service and communicated to egress nodes in the response to a successful get-new-egress-key rpc. Egress keys SHOULD be based on a random value and unique per new key requested.¶
Egress nodes provide their egress key when performing group management functions (join and leave operations).¶
TBD: better explanation about how the service times out egress nodes that don't refresh their egress key on schedule, and how egress nodes that reconnect can attempt to refresh the prior key they were using, but must request a new one on error. Probably define a state per egress key (e.g. active vs. recently expired vs. non-existant) for the MNAT service to maintain. Explain how the MNAT service should use population count from the egress joins to make prioritization decisions for the assignment of flows when there is limited flow space. Probably reference CBACC in that explanation (I-D.draft-ietf-mboned-cbacc).¶
The join-global and leave-global RPCs in the YANG model provide a mechanism for egress nodes to directly advertise their group membership to the MNAT service for externally addressed (S,G)s.¶
Egress nodes advertise their group membership to external (S,G)s to the MNAT service and also advertise group membership to their next-hop router using IGMP or MLD for the locally mapped addressing withing the network. Joins and leaves for the locally mapped network addresses occur in response to downstream joins for an external (S,G) that has or gains a mapping according to the MNAT service, when the join or leave propagates to the egress node.¶
Payloads of the locally mapped traffic should be treated as though they were carried in packets addressed as the external (S,G), including any authentication checks that should be performed for the traffic. Egress nodes that forward traffic (non-virtual egress nodes) will perform an address translation from the locally mapped addreessing to the original (S,G) (according to the address mapping the MNAT service provides) before forwarding packets matching a locally mapped address. It is the responsibility of the MNAT service and the network that operates it to ensure that multiple different traffic streams are not merged to the same locally mapped addresses in a way that collides.¶
Like egress nodes, ingress nodes monitor the assignments provided by the MNAT service and perform network address translation and group membership propagation. Ingress nodes perform the translation from an external (S,G) to the internally mapped addressing for the local network transport.¶
In general, ingress nodes are translating traffic before the in-network multicast fanout to multiple egress nodes. So an ingress node is generally assumed to be feeding one or more egress nodes. Because one ingress node can feed many egress nodes, ingress nodes should be given priority ahead of egress nodes for notifications about changes to the address mapping from the MNAT service.¶
The details of the address assignment strategies used by the internal logic of the MNAT service are out of scope for this document. Different instances of MNAT services are expected to use a wide range of considerations specific to the networks in which the instances operate.¶
However, outside of address assignment there are some operational points an MNAT service instance should take into consideration:¶
Assignment Transition Grace Period¶
It's recommended to provide a grace period between reassigning a local address mapping to a new external (S,G) after unassigning its mapping to an old (S,G). The grace period should account for the expected time for the connected ingress and egress nodes to process the unassigning of the external (S,G) and for egress nodes to perform leave operations for the old locally mapped address, and for the leave operations to propagate through the network.¶
Scaling¶
The MNAT service should be appropriately provisioned to support the expected number of ingress and egress nodes within the network. In an eyeball network, restrictions on the number of egress nodes per shared receiver IP address may be appropriate, to avoid a rogue client application from forming an excessive number of egress connections. Alternately, for bump-in-the-wire deployments of egress nodes in CPE devices it may be appropriate to authenticate the egress connections with a client certificate for each home to avoid denial of service attacks based on overloading the MNAT service with egress connections.¶
Additionally, it's RECOMMENDED to provide per-egress limits on the number of external simultaneous (S,G)s permitted per egress at a level appropriate to the scaling limitations for the network, to avoid denial of service attacks based on overloading the group assignments.¶
TBD: show what an expected example message sequence or 2 would look like.¶
The tree diagram below uses the notation defined in [RFC8340].¶
module: ietf-mnat
+--ro assigned-channels
+--ro mapped-sg* [id]
+--ro id assignment-id
+--ro state assignment-state
+--ro global-subscription
| +--ro (channel-type)?
| +--:(ssm-channel)
| | +--ro source inet:ip-address
| | +--ro group
| | rt-types:ip-multicast-group-address
| +--:(asm-channel)
| +--ro asm-group
| rt-types:ip-multicast-group-address
+--ro local-mapping
+--ro (mapping-type)?
+--:(local-multicast-mapping)
+--ro (channel-type)?
+--:(ssm-channel)
| +--ro source inet:ip-address
| +--ro group
| rt-types:ip-multicast-group-address
+--:(asm-channel)
+--ro asm-group
rt-types:ip-multicast-group-address
rpcs:
+---x get-new-egress-key
| +--ro output
| +--ro egress-id egress-key
| +--ro refresh-period? uint16
+---x refresh-egress-key
| +---w input
| +---w egress-id egress-key
+---x join-global
| +---w input
| +---w egress-id egress-key
| +---w (channel-type)?
| +--:(ssm-channel)
| | +---w source inet:ip-address
| | +---w group
| | rt-types:ip-multicast-group-address
| +--:(asm-channel)
| +---w asm-group
| rt-types:ip-multicast-group-address
+---x leave-global
+---w input
+---w egress-id egress-key
+---w (channel-type)?
+--:(ssm-channel)
| +---w source inet:ip-address
| +---w group
| rt-types:ip-multicast-group-address
+--:(asm-channel)
+---w asm-group
rt-types:ip-multicast-group-address
<CODE BEGINS> file ietf-mnat@2020-11-02.yang
module ietf-mnat {
yang-version 1.1;
namespace "urn:ietf:params:xml:ns:yang:ietf-mnat";
prefix mnat;
import ietf-yang-types {
prefix yang;
}
import ietf-inet-types {
prefix inet;
reference
"RFC 6991: Common YANG Data Types";
}
import ietf-routing-types {
prefix "rt-types";
reference "RFC 8294";
}
organization
"IETF MBONED (Multicast Backbone Deployment) Working Group";
contact
"WG Web: <https://datatracker.ietf.org/wg/mboned/>
WG List: <mailto:mboned@ietf.org>
Author: Jake Holland
<mailto:jakeholland.net@gmail.com>";
description
"Multicast Network Address Translation Model.
Copyright (c) 2012 - 2020 IETF Trust and the persons
identified as authors of the code. All rights reserved.
Redistribution and use in source and binary forms, with or
without modification, is permitted pursuant to, and subject
to the license terms contained in, the Simplified BSD
License set forth in Section 4.c of the IETF Trust's
Legal Provisions Relating to IETF Documents
(https://trustee.ietf.org/license-info).
This version of this YANG module is part of RFC XXXX; see
the RFC itself for full legal notices.";
revision "2020-10-22" {
description
"Initial version.";
}
grouping multicast-channel {
choice channel-type {
description
"ASM or SSM multicast channels can be represented.";
case ssm-channel {
leaf source {
type inet:ip-address;
mandatory true;
description
"Source address of a multicast channel";
}
leaf group {
type rt-types:ip-multicast-group-address;
mandatory true;
description "The global (S,G)'s group address";
}
}
case asm-channel {
leaf asm-group {
type rt-types:ip-multicast-group-address;
mandatory true;
description "The global (S,G)'s group address";
}
}
}
}
typedef egress-key {
type string;
description
"A key for egress identification.";
}
typedef assignment-id {
type uint32;
description
"A type for assignment identifiers.";
}
identity assignment-state {
description
"Base identity to represent assignment states";
}
typedef assignment-state {
type identityref {
base assignment-state;
}
description "Status of an assigned (S,G).";
}
identity unassigned {
base assignment-state;
description
"Represents an unassigned global (S,G) that cannot be
received in the local network.";
}
identity assigned-local-multicast {
base assignment-state;
description
"Represents an assigned global (S,G) that can be
received in the local network by joining the associated
local-mapping.";
}
container assigned-channels {
config false;
description
"MNAT mappings of global (S,G)s into a local transport.";
list mapped-sg {
key "id";
description
"The local network's assignment of global channels to
local transport characteristics.";
leaf id {
type assignment-id;
mandatory true;
description
"Identifier for this assignment.";
}
leaf state {
type assignment-state;
mandatory true;
description
"Status of the global (S,G)s that are assigned in the
local network.";
}
container global-subscription {
description
"The global channel that's mapped.";
uses multicast-channel;
}
container local-mapping {
choice mapping-type {
description
"The description of how the global channel is
transported within the local network";
case local-multicast-mapping {
description
"Defines the use of a local multicast (S,G) or
(*,G).";
uses multicast-channel;
}
}
}
}
}
rpc get-new-egress-key {
description
"Obtain a secret key unique to an individual mnat-egress
instance, assigned by the server and used for subscription
management.";
output {
leaf egress-id {
type egress-key;
mandatory true;
description
"Egress identifier for subscription management.";
}
leaf refresh-period {
type uint16;
default 10;
description
"Number of seconds to wait between refresh messages.";
}
}
}
rpc refresh-egress-key {
description
"A secret key unique to an individual mnat-egress instance,
assigned by the server and used for subscription
management.";
input {
leaf egress-id {
type egress-key;
mandatory true;
description
"Egress identifier for subscription management.";
}
}
}
rpc join-global {
description
"An mnat-egress instance calls this RPC to register to the
network an interest in a global (S,G). Depending on
popularity and local network decisions, this may result in
adding and possibly removing an entry in
assigned-channels/mapped-sg.";
input {
leaf egress-id {
type egress-key;
mandatory true;
description
"Egress identifier.";
}
uses multicast-channel;
}
}
rpc leave-global {
description
"An mnat-egress instance calls this RPC to register to the
network an interest in a global (S,G). Depending on
popularity and local network decisions, this may result in
adding and possibly removing an entry in
assigned-channels/mapped-sg.";
input {
leaf egress-id {
type egress-key;
mandatory true;
description
"Egress identifier.";
}
uses multicast-channel;
}
}
}
<CODE ENDS>¶
This document adds one YANG module to the "YANG Module Names" registry maintained at <https://www.iana.org/assignments/yang-parameters>. The following registrations are made, per the format in Section 14 of [RFC6020]:¶
name: ietf-mnat
namespace: urn:ietf:params:xml:ns:yang:ietf-mnat
prefix: mnat
reference: I-D.draft-jholland-mboned-mnat
¶
This document adds the following registration to the "ns" subregistry of the "IETF XML Registry" defined in [RFC3688], referencing this document.¶
URI: urn:ietf:params:xml:ns:yang:ietf-mnat
Registrant Contact: The IESG.
XML: N/A, the requested URI is an XML namespace.
¶
This document adds one service name to the "Service Name and Transport Protocol Port Number Registry" maintained at <https://www.iana.org/assignments/service-names-port-numbers>. The following registrations are made, per the format in Section 8.1.1 of [RFC6335]:¶
Service Name: mnat
Transport Protocol(s): TCP, UDP
Assignee: IESG <iesg@ietf.org>
Contact: IETF Chair <chair@ietf.org>
Description: The MNAT service (RESTCONF that
includes ietf-mnat YANG model)
Reference: I-D.draft-jholland-mboned-mnat
Port Number: N/A
Service Code: N/A
Known Unauthorized Uses: N/A
Assignment Notes: N/A
¶
TBD. (What, me worry?)¶
Notable points to cover:¶
Many thanks to anyone who reads this and provides feedback.¶