| Homenet Working Group | M. Stenberg |
| Internet-Draft | |
| Intended status: Standards Track | S. Barth |
| Expires: April 30, 2015 | |
| October 27, 2014 |
Home Networking Control Protocol
draft-ietf-homenet-hncp-02
This document describes the Home Networking Control Protocol (HNCP), a minimalist state synchronization protocol for homenet routers.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on April 30, 2015.
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.
HNCP is designed to synchronize state across a homenet (or other small site) in order to facilitate automated configuration within the site. The design supports border discovery, IP prefix distribution [I-D.ietf-homenet-prefix-assignment], and service discovery across multiple links[I-D.stenberg-homenet-dnssd-hybrid-proxy-zeroconf].
HNCP is designed to provide enough information for a routing protocol to operate without homenet-specific extensions. In homenet environments where multiple IPv6 prefixes are present, routing based on source and destination address is necessary [I-D.troan-homenet-sadr]. Routing protocol requirements for source and destination routing are described in section 3 of [I-D.baker-rtgwg-src-dst-routing-use-cases].
A GPLv2-licensed implementation of the HNCP protocol is currently under development at https://github.com/sbyx/hnetd/ and the binaries are available in the routing feed of OpenWrt trunk release. Some information how to get started with it is available at http://www.homewrt.org/doku.php?id=run-conf. Comments and/or pull requests are welcome.
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].
The data model of the HNCP protocol is simple: Every participating node has (and also knows for every other participating node):
HNCP is designed to run on UDP port IANA-UDP-PORT, using both link-local scoped IPv6 unicast and link-local scoped IPv6 multicast messages to address IANA-MULTICAST-ADDRESS for transport. The protocol consists of Trickle [RFC6206] driven multicast status messages to indicate changes in shared TLV data, and unicast state synchronization message exchanges when the Trickle state is found to be inconsistent.
Each node MUST send link-local multicast NetState Messages [net-message] each time the Trickle algorithm [RFC6206] indicates they should on each link the protocol is active on. When the locally stored network state hash changes (either by a local node event that affects the TLV data, or upon receipt of more recent data from another node), all Trickle instances MUST be reset. Trickle state MUST be maintained separately for each link.
Trickle algorithm has 3 parameters; Imin, Imax and k. Imin and Imax represent minimum and maximum values for I, which is the time interval during which at least k Trickle updates must be seen on a link to prevent local state transmission. Bounds for recommended Trickle values are described below.
Protocol messages are encoded as purely as a sequence of TLV objects [tlvs]. This section describes which set of TLVs MUST or MAY be present in a given message.
In order to facilitate fast comparing of local state with that in a received message update, all TLVs in every encoding scope (either root level, within the message itself, or within a container TLV) MUST be placed in ascending order based on the binary comparison of both TLV header and value. By design, the TLVs which MUST be present have the lowest available type values, ensuring they will naturally occur at the start of the Protocol Message, resembling a fixed format preamble.
This Message SHOULD be sent as a multicast message.
The following TLVs MUST be present at the start of the message:
The NetState Message MAY contain Node State TLV(s) [node-state]. If so, either all Node State TLVs are included (referred to as a "long" NetState Message), or none are included (referred to as a "short" NetState Message). The NetState Message MUST NOT contain only a portion of Node State TLVs as this could cause problems with the Protocol Message Processing [reception] algorithm. Finally, if the long version of the NetState message would exceed the minimum IPv6 MTU when sent, the short version of the NetState message MUST be used instead.
This Message MUST be sent as a unicast message.
The following TLVs MUST be present at the start of the message:
This Message MUST be sent as a unicast message.
MUST be present:
This Message MUST be sent as a unicast message.
MUST be present:
The majority of status updates among known nodes are handled via the Trickle-driven updates [trickle-updates]. This section describes processing of messages as received, along with associated actions or responses.
HNCP is designed to operate between directly connected neighbors on a shared link using link-local IPv6 addresses. If the source address of a received HNCP packet is not an IPv6 link-local unicast address, the packet SHOULD be dropped. Similarly, if the destination address is not IPv6 link-local unicast or IPv6 link-local multicast address, packet SHOULD be dropped.
Upon receipt of:
Each node is responsible for publishing a valid set of data TLVs. When there is a change in a node's set of data TLVs, the update number MUST be incremented accordingly.
If a message containing Node State TLVs [node-state] is received via unicast or multicast with the node's own node identifier and a higher update number than current local value, or the same update number and different hash, there is an error somewhere. A recommended default way to handle this is to attempt to assert local state by increasing the local update number to a value higher than that received and republish node data using the same node identifier. If this happens more than 3 times in 60 seconds and the local node identifier is not globally unique, there may be more than one router with the same node identifier on the network. If there is no global guarantee of unique node identifier, a new node identifier SHOULD be generated and node data republished accordingly.
In all cases, if node data for any node changes, all Trickle instances MUST be considered inconsistent (I=Imin + timer reset).
Whenever multicast message or unicast reply is received on a link from another node, the node should be added as Neighbor TLV [neighbor-tlv] for current node. If nothing (for example - no router advertisements, no HNCP traffic) is received from that neighbor in Imax seconds and the neighbor is not in neighbor discovery cache, and no layer 2 indication of presence is available, at least 3 attempts to ping it with request network state message [req-net-message] SHOULD be sent with increasing timeouts (e.g. 1, 2, 4 seconds). If even after suitable period after the last message nothing is received, the Neighbor TLV MUST be removed so that there are no dangling neighbors. As an alternative, if there is a layer 2 unreachability notification of some sort available for either whole link or for individual neighbor, it MAY be used to immediately trigger removal of corresponding Neighbor TLV(s).
When node data has changed, the neighbor graph SHOULD be traversed for each node following the bidirectional neighbor relationships. These are identified by looking for neighbor TLVs on both nodes, that have the remote node's identifier hash as h(neighbor node identifier), and local and neighbor link identifiers swapped. After the traverse, unreachable nodes SHOULD be purged after some grace period. During the grace period, the unreachable nodes MUST NOT be used for calculation of network state hash, or even be provided to any applications that need to use the whole TLV graph.
Every TLV is encoded as 2 octet type, followed by 2 octet length (of the whole TLV, including header; 4 means no value), and then the value itself (if any). The actual length of TLV MUST be always divisible by 4; if the length of the value is not, zeroed padding bytes MUST be inserted at the end of TLV. The padding bytes MUST NOT be included in the length field.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type | Length |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Value |
| (variable # of bytes) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Encoding of type=123 (0x7b) TLV with value 'x' (120 = 0x78): 007B 0005 7800 0000
Notation:
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: REQ-NETWORK-STATE (2) | Length: 4 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: REQ-NODE-DATA (3) | Length: 20 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| H(node identifier) |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: NODE-LINK (1) | Length: 24 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| H(node identifier) |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Link-Identifier |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: NETWORK-STATE (4) | Length: 20 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| H(H(node data TLV 1) .. H(node data TLV N)) |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The Node Data TLVs are ordered for hashing by octet comparison of the corresponding node identifier hashes in ascending order.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: NODE-STATE (5) | Length: 44 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| H(node identifier) |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Update Sequence Number |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Milliseconds since Origination |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| H(node data TLV) |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The whole network should have roughly the same idea about the time since origination, i.e. even the originating router should increment the time whenever it needs to send a new Node State TLV regarding itself without changing the corresponding Node Data TLV. This age value is not included within the Node Data TLV, however, as that is immutable and potentially signed by the originating node at the time of origination.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: NODE-DATA (6) | Length: >= 24 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| H(node identifier) |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Update Sequence Number |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Nested TLVs containing node information |
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: NEIGHBOR (8) | Length: 28 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| H(neighbor node identifier) |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Neighbor Link Identifier |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Local Link Identifier |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
This TLV indicates that the node in question vouches that the specified neighbor is reachable by it on the local link id given. This reachability may be unidirectional (if no unicast exchanges have been performed with the neighbor). The presence of this TLV at least guarantees that the node publishing it has received traffic from the neighbor recently. For guaranteed bidirectional reachability, existence of both nodes' matching Neighbor TLVs should be checked.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: CUSTOM-DATA (9) | Length: >= 12 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| H-64(URI) |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Opaque Data |
This TLV can be used to contain anything; the URI used should be under control of the author of that specification. For example:
V=H-64('http://example.com/author/json-for-hncp') .. '{"cool": "json extension!"}'
or
V=H-64('mailto:author@example.com') .. '{"cool": "json extension!"}'
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: VERSION (10) | Length: >= 8 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Version |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| User-agent |
This TLV indicates which version of HNCP TLV binary structures is in use by this particular node. All TLVs within node data from nodes that do not publish version TLV, or with different Version value than locally supported one MUST be ignored (but forwarded). The user-agent is an optional human-readable UTF-8 string that can describe e.g. current HNCP implementation version. This draft describes Version=1 TLVs.
This section defines border discovery algorithm derived from the edge router interactions described in the Basic Requirements for IPv6 Customer Edge Routers [RFC7084]. The algorithm is designed to work for both IPv4 and IPv6 (single or dual-stack).
In order to avoid conflicts between border discovery and homenet routers running DHCP [RFC2131] or DHCPv6-PD [RFC3633] servers each router MUST implement the following mechanism based on The User Class Option for DHCP [RFC3004] or its DHCPv6 counterpart [RFC3315] respectively into its DHCP and DHCPv6-logic:
The border discovery auto-detection algorithm works as follows, with evaluation stopping at first match:
A router MUST allow setting a category of either auto-detected, internal or external for each interface which is suitable for both internal and external connections. In addition the following specializations of the internal category are defined to modify the local router behavior:
A homenet router SHOULD provide basic connectivity to legacy CERs [RFC7084] connected to internal interfaces in order to allow coexistence with existing devices.
Each router MUST continuously scan each active interface that does not have a fixed category in order to dynamically reclassify it if necessary. The router therefore runs an appropriately configured DHCP and DHCPv6-client as long as the interface is active including states where it considers the interface to be internal. The router SHOULD wait for a reasonable time period (5 seconds as a possible default) in which the DHCP-clients can acquire a lease before treating a newly activated or previously external interface as internal. Once it treats a certain interface as internal it MUST start forwarding traffic with appropriate source addresses between its internal interfaces and allow internal traffic to reach external networks. Once a router detects an interface to be external it MUST stop any previously enabled internal forwarding. In addition it SHOULD announce the acquired information for use in the homenet as described in later sections of this draft if the interface appears to be connected to an external network.
To distribute an external connection in the homenet an edge router announces one or more delegated prefixes and associated DHCP(v6)-encoded auxiliary information like recursive DNS-servers. Each external connection is announced using one container-TLV as follows:
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type: EXTERNAL-CONNECTION (41)| Length: > 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Nested TLVs |
Auxiliary connectivity information is encoded as a stream of DHCPv6-attributes or DHCP-attributes placed inside a TLV of type EXTERNAL-CONNECTION or DELEGATED-PREFIX (for IPv6 prefix-specific information). There MUST NOT be more than one instance of this TLV inside a container and the order of the DHCP(v6)-attributes contained within it MUST be preserved as long as the information contained does not change. The TLVs are encoded as follows:
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type: DHCPV6-DATA (45) | Length: > 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | DHCPv6 attribute stream |
and
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type: DHCP-DATA (44) | Length: > 4 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | DHCP attribute stream |
Each delegated prefix is encoded using one TLV inside an EXTERNAL-CONNECTION TLV. For external IPv4 connections the prefix is encoded in the form of an IPv4-mapped address [RFC4291] and is usually from a private address range [RFC1918]. The related TLV is defined as follows.
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type: DELEGATED-PREFIX (42) | Length: >= 13 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Valid until (milliseconds) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Preferred until (milliseconds) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Prefix Length | | +-+-+-+-+-+-+-+-+ Prefix Address [+ nested TLVs] + | |
In order for routers to use the distributed information, prefixes and addresses have to be assigned to the interior links of the homenet. A router MUST therefore implement the algorithm defined in Prefix and Address Assignment in a Home Network [I-D.ietf-homenet-prefix-assignment]. In order to announce the assigned prefixes the following TLVs are defined.
Each assigned prefix is given to an interior link and is encoded using one TLVs. Assigned IPv4 prefixes are stored as mapped IPv4-addresses. The TLV is defined as follows:
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type: ASSIGNED-PREFIX (43) | Length: >= 9 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Link Identifier | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | R. |A| Pref. | Prefix Length | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Prefix Address + | |
In some cases (e.g. IPv4) the set of addresses is very limited and stateless mechanisms are not really suitable for address assignment. Therefore HNCP can manage router address in these cases by itself. Each router assigning an address to one of its interfaces announces one TLV of the following kind:
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type: ROUTER-ADDRESS (46) | Length: 24 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Link Identifier | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | | | Router Address | | | | | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Service discovery is generally limited to a local link. [I-D.stenberg-homenet-dnssd-hybrid-proxy-zeroconf] defines a mechanism to automatically extended DNS-based service discovery across multiple links within the home automatically. Following TLVs MAY be used to provide transport for that specification.
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: DNS-DELEGATED-ZONE (50) | Length: >= 21 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| Address |
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Reserved |S|B| |
+-+-+-+-+-+-+-+-+ Zone (DNS label sequence - variable length) |
| |
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: DOMAIN-NAME (51) | Length: >= 4 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Domain (DNS label sequence - variable length) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type: ROUTER-NAME (52) | Length: >= 4 |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Name (not null-terminated - variable length) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
In order to be advertised for use within the homenet, a routing protocol MUST:
A router MUST NOT announce that it supports a certain routing protocol if its implementation of the routing protocol does not meet these requirements, e.g. it does not implement extensions that are necessary for compliance.
Each router SHOULD announce all routing protocols that it is capable of supporting in the homenet. It MUST announce at least one protocol or the fallback mechanism to be considered for protocol selection and MAY omit announcing the fallback mechanism if it announces at least one other routing protocol. It SHOULD assign a preference value for each protocol that indicates its desire to use said protocol over other protocols it supports and SHOULD make these values configurable.
Each router includes one HNCP TLV of type ROUTING-PROTOCOL for every such routing protocol. This TLV is defined as follows:
0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type: ROUTING-PROTOCOL (60) | Length: 6 | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Protocol ID | Preference | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
When HNCP detects that a device has joined or left the homenet it MUST examine all advertised routing protocols and preference values from all devices announcing at least one ROUTING-PROTOCOL-TLV in order to find the one routing protocol which:
If the router protocol selection results in the need to change from one routing protocol to another on the homenet, the router MUST stop the previously running protocol, remove associated routes, and start the new protocol in a graceful manner. If there is no common routing protocol available among all homenet routers, routers MUST utilize the Fallback Mechanism [fallback].
In cases where there is no commonly supported routing protocol available the following fallback algorithm is run to setup routing and preserve interoperability among the homenet. While not intended to replace a routing protocol, this mechanism provides a valid - but not necessarily optimal - routing topology. This algorithm uses the node and neighbor state already synchronized by HNCP, and therefore does not require any additional protocol message exchange.
General security issues for home networks are discussed at length in [RFC7368]. The protocols used to set up IP in home networks today have rarely security enabled within the control protocol itself. For example, DHCP has defined [RFC3118] to authenticate DHCP messages, but this is very rarely implemented in large or small networks. Further, while PPP can provide secure authentication of both sides of a point to point link, it is most often deployed with one-way authentication of the subscriber to the ISP, not the ISP to the subscriber.
HNCP itself sends messages as clear text which is as secure, or insecure, as the security of the link it runs on. As HNCP messages are sent over IPv6 UDP, IPsec may be used for confidentiality or message authentication. This requires manually keyed IPsec per-port granularity for port IANA-UDP-PORT UDP traffic, and a pre-shared key has to be utilized in this case given IKE cannot be used with multicast traffic. This seems acceptable, though, as most routing protocols also operate based on pre-shared keys, and the homenet architecture draft explicitly allows their use for securing them[RFC7368]. Other traffic security mechanisms are out of scope for this specification. There is ongoing work [I-D.barth-homenet-hncp-security-trust] to define a mechanism that can be used with HNCP and be more user-friendly than pre-shared keys.
IANA should set up a registry (policy TBD) for HNCP TLV types, with following initial contents:
0: Reserved (should not happen on wire)
1: Node link
2: Request network state
3: Request node data
4: Network state
5: Node state
6: Node data
7: (unused - was node public key, but never implemented)
8: Neighbor
9: Custom
10: Version
41: External connection
42: Delegated prefix
43: Assigned prefix
44: DHCP-data
45: DHCPv6-data
46: Router-address
50: DNS Delegated Zone
51: Domain name
52: Node name
60: Routing protocol
65535: Signature
HNCP will also require allocation of a UDP port number IANA-UDP-PORT, as well as IPv6 link-local multicast address IANA-MULTICAST-ADDRESS.
| [RFC2119] | Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. |
| [RFC6206] | Levis, P., Clausen, T., Hui, J., Gnawali, O. and J. Ko, "The Trickle Algorithm", RFC 6206, March 2011. |
| [I-D.ietf-homenet-prefix-assignment] | Pfister, P., Paterson, B. and J. Arkko, "Prefix and Address Assignment in a Home Network", Internet-Draft draft-ietf-homenet-prefix-assignment-01, October 2014. |
| [I-D.stenberg-homenet-dnssd-hybrid-proxy-zeroconf] | Stenberg, M., "Auto-Configuration of a Network of Hybrid Unicast/Multicast DNS-Based Service Discovery Proxy Nodes", Internet-Draft draft-stenberg-homenet-dnssd-hybrid-proxy-zeroconf-01, June 2014. |
Should we use MD5 hashes, or EUI-64 node identifier to identify nodes?
Is there a case for non-link-local unicast? Currently explicitly stating this is link-local only protocol.
Consider if using Trickle with k=1 really pays off, as we need to do reachability checks if layer 2 does not provide them periodically in any case. Using Trickle with k=inf would remove the need for unicast reachability checks, but at cost of extra multicast traffic. On the other hand, N*(N-1)/2 unicast reachability checks when lot of routers share a link is not appealing either.
Valid and preferred are now 32 bit millisecond and you cannot even represent a month in them; is this enough? Or should we switch to 32 bit seconds (or 64 bit milliseconds)?
Q: Why not use TCP?
A: It does not address the node discovery problem. It also leads to N*(N-1)/2 connections when N nodes share a link, which is awkward.
Q: Why not multicast-only?
A: It would require defining application level fragmentation scheme. Hopefully the data amounts used will stay small so we just trust unicast UDP to handle 'big enough' packets to contain single node's TLV data. On some link layers unicast is also much more reliable than multicast, especially for large packets. Also on wireless, multicast is much more power expensive than unicast.
Q: Why so long IDs?
A: Scalability of protocol is not really affected by using real (=cryptographic) hash function.
Q: Why trust IPv6 fragmentation in unicast case? Why not do L7 fragmentation?
A: Because it will be there for a while at least. And while PMTU et al may be problems on open internet, in a home network environment UDP fragmentation should NOT be broken in the foreseeable future.
Q: Should there be nested container syntax that is actually self-describing? (i.e. type flag that indicates container, no body except sub-TLVs?)
A: Not for now, but perhaps valid design.. TBD.
Q: Why not doing (performance thing X, Y or Z)?
A: This is designed mostly to be minimal (only timers Trickle ones; everything triggered by Trickle-driven messages or local state changes). However, feel free to suggest better (even more minimal) design which works.
draft-ietf-homenet-hncp-02: Removed any built-in security. Relying on IPsec. Reorganized interface categories, added requirements languages, made manual border configuration a MUST-support. Redesigned routing protocol election to consider non-router devices.
draft-ietf-homenet-hncp-01: Added (MAY) guest, ad-hoc, hybrid categories for interfaces. Removed old hnetv2 reference, and now pointing just to OpenWrt + github. Fixed synchronization algorithm to spread also same update number, but different data hash case. Made purge step require bidirectional connectivity between nodes when traversing the graph. Edited few other things to be hopefully slightly clearer without changing their meaning.
draft-ietf-homenet-hncp-00: Added version TLV to allow for TLV content changes pre-RFC without changing IDs. Added link id to assigned address TLV.
As usual, this draft is available at https://github.com/fingon/ietf-drafts/ in source format (with nice Makefile too). Feel free to send comments and/or pull requests if and when you have changes to it!
Thanks to Ole Troan, Pierre Pfister, Mark Baugher, Mark Townsley and Juliusz Chroboczek for their contributions to the draft.
Thanks to Eric Kline for the original border discovery work.