| ROLL Working Group | M. Robles |
| Internet-Draft | Ericsson |
| Intended status: Informational | M. Richardson |
| Expires: January 21, 2017 | SSW |
| P. Thubert | |
| Cisco | |
| July 20, 2016 |
When to use RFC 6553, 6554 and IPv6-in-IPv6
draft-ietf-roll-useofrplinfo-07
This document looks at different data flows through LLN (Low-Power and Lossy Networks) where RPL (IPv6 Routing Protocol for Low-Power and Lossy Networks) is used to establish routing. The document enumerates the cases where RFC 6553, RFC 6554 and IPv6-in-IPv6 encapsulation is required. This analysis provides the basis on which to design efficient compression of these headers.
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 January 21, 2017.
Copyright (c) 2016 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.
RPL (IPv6 Routing Protocol for Low-Power and Lossy Networks) [RFC6550] is a routing protocol for constrained networks. RFC 6553 [RFC6553] defines the "RPL option" (RPI), carried within the IPv6 Hop-by-Hop header to quickly identify inconsistencies (loops) in the routing topology. RFC 6554 [RFC6554] defines the "RPL Source Route Header" (RH3), an IPv6 Extension Header to deliver datagrams within a RPL routing domain, particularly in non-storing mode.
These various items are referred to as RPL artifacts, and they are seen on all of the data-plane traffic that occurs in RPL routed networks; they do not in general appear on the RPL control plane traffic at all which is mostly hop-by-hop traffic (one exception being DAO messages in non-storing mode).
It has become clear from attempts to do multi-vendor interoperability, and from a desire to compress as many of the above artifacts as possible that not all implementors agree when artifacts are necessary, or when they can be safely omitted, or removed.
An interim meeting went through the 24 cases defined here to discover if there were any shortcuts, and this document is the result of that discussion. This document should not be defining anything new, but it may clarify what is correct and incorrect behaviour.
The related document A Routing Header Dispatch for 6LoWPAN (6LoRH) [I-D.ietf-roll-routing-dispatch] defines a method to compress RPL Option information and Routing Header type 3 [RFC6554], an efficient IP-in-IP technique, and use cases proposed for the [Second6TischPlugtest] involving 6loRH.
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].
Terminology defined in [RFC7102] applies to this document: LBR, LLN, RPL, RPL Domain and ROLL.
The term "hop-by-hop IPv6-in-IPv6" header refers to: adding a header that originates from a node to an adjacent node, using the addresses (usually the GUA or ULA, but could use the link-local addresses) of each node. If the packet must traverse multiple hops, then it must be decapsulated at each hop, and then re-encapsulated again in a similar fashion.
A RPL network is composed of a 6LBR (6LoWPAN Border Router), Backbone Router (6BBR), 6LR (6LoWPAN Router) and 6LN (6LoWPAN Node) as leaf logically organized in a DODAG structure (Destination Oriented Directed Acyclic Graph).
RPL defines the RPL Control messages (control plane), a new ICMPv6 [RFC4443] message with Type 155. DIS (DODAG Information Solicitation), DIO (DODAG Information Object) and DAO (Destination Advertisement Object) messages are all RPL Control messages but with different Code values. A RPL Stack is showed in Figure 1.
RPL supports two modes of Downward traffic: in storing mode (RPL-SM), it is fully stateful or an in non-storing (RPL-NSM), it is fully source routed. A RPL Instance is either fully storing or fully non-storing, i.e. a RPL Instance with a combination of storing and non-storing nodes is not supported with the current specifications at the time of writing this document.
+--------------+
| Upper Layers |
| |
+--------------+
| RPL |
| |
+--------------+
| ICMPv6 |
| |
+--------------+
| IPv6 |
| |
+--------------+
| 6LoWPAN |
| |
+--------------+
| PHY-MAC |
| |
+--------------+
Figure 1: RPL Stack.
+---------+
+---+Internet |
| +---------+
|
+----+--+
| DODAG | node:01
+---------+ Root +----------+
| | 6LBR | |
| +----+--+ |
| | |
| | |
... ... ...
| | |
+-----+-+ +--+---+ +--+---+
|6LR | | | | |
+-----+ | | | | |
| | 11 | | 12 | | 13 +------+
| +-----+-+ +-+----+ +-+----+ |
| | | | |
| | | | |
| 21 | 22 | 23 | 24 | 25
+-+---+ +-+---+ +--+--+ +- --+ +---+-+
|Leaf | | | | | |Leaf| |Leaf |
| 6LN | | | | | | 6LN| | 6LN |
+-----+ +-----+ +-----+ +----+ +-----+
Figure 2: A reference RPL Topology.
In Figure 2 is showed the reference RPL Topology for this document. The numbers in or above the nodes are there so that they may be referenced in subsequent sections. In the figure, a 6LN can be a router or a host. The 6LN leafs marked as (21) and (25) are routers. The leaf marked 6LN (24) is a device which does not speak RPL at all (not-RPL-aware), but uses Router-Advertisements, 6LowPAN DAR/DAC and efficient-ND only to participate in the network [RFC6775]. In the document this leaf (24) is often named IPv6 node. The 6LBR in the figure is the root of the Global DODAG.
This document is in part motivated by the work that is ongoing at the 6TiSCH working group. The 6TiSCH architecture [I-D.ietf-6tisch-architecture] draft explains the network architecture of a 6TiSCH network.
In data plane context a combination of RFC6553, RFC6554 and IPv6-in-IPv6 encapsulation is going to be analyzed for the following traffic flows.
This version of the document assumes the changes in [I-D.ietf-6man-rfc2460bis] are passed.
This document assumes the rule that a Header cannot be inserted or removed on the fly inside an IPv6 packet that is being routed. This is a fundamental precept of the IPv6 architecture as outlined in [RFC2460] is that Extensions may not be added or removed except by the sender or the receiver. (A revision to RFC2460 considered changing this rule, but has kept it)
But, options in the Hop-by-Hop option which are marked with option type 01 ([RFC2460] section 4.2 and [I-D.ietf-6man-rfc2460bis]) SHOULD be ignored when received by a host or router which does not understand that option.
This means that in general, any packet that leaves the RPL domain of an LLN (or leaves the LLN entirely) will NOT be discarded, even if it has the [RFC6553] RPL Option Header known as the RPI or [RFC6554] SRH3 Extension Header (S)RH3.
With abolition of one of these rules it means that the RPI Hop-by-Hop option MAY be left in place even if the end host does host understand it. This collapses many of the cases above (where it says "or")
An intermediate router that needs to add an extension header (SHR3 or RPI Option) must encapsulate the packet in an (additional) outer IP header where the new header can be placed.
This also means that a Header can only be removed by an intermediate router if it is placed in an encapsulating IPv6 Header, and in that case, the whole encapsulating header must be removed - a replacement may be added. Further, an intermediate router can only remove such an outer header if that outer header has the router as the destination!
Both RPI and RH3 headers may be modified by routers on the path of the packet without the need to add to remove an encapsulating header. Both headers were designed with this modification in mind, and both the RPL RH and the RPL option are marked mutable but recoverable, so an IPsec AH security header can be applied across these headers, but it may not secure all the values in those headers.
RPI should be present in every single RPL data packet. There is one exception in non-storing mode: when a packet is going down from the root. In a downward non-storing mode, the entire route is written, so there can be no loops by construction, nor any confusion about which forwarding table to use. There may be cases (such as in 6tisch) where the instanceID may still be needed to pick an appropriate priority or channel at each hop.
In the tables present in this document, the term "RPL aware leaf" is has been shortened to "Raf", and "not-RPL aware leaf" has been shortened to "~Raf" to make the table fit in available space.
The earlier examples are more extensive to make sure that the process is clear, while later examples are more consise.
In storing mode (fully stateful), the sender cannot determine whether the destination is RPL-capable and thus would need an IP-in-IP header. The IP-in-IP header needs to be addressed on a hop-by-hop basis so that the last 6LR can remove the RPI header. Additionally, The sender can determine if the destination is inside the LLN by looking if the destination address is matched by the DIO's PIO option.
The following table summarizes what headers are needed in the following scenarios, and indicates when the IP-in-IP header must be inserted on a hop-by-hop basis, and when it can target the destination node directly. There are three possible situations: hop-by-hop necessary (indicated by "hop"), or destination address possible (indicated by "dst"). In all cases hop by hop can be used. In cases where no IP-in-IP header is needed, the column is left blank.
The leaf can be a router 6LR or a host, both indicated as 6LN.
| Use Case | RPI | RH3 | IP-in-IP | IP-in-IP dst |
|---|---|---|---|---|
| Raf to root | Yes | No | No | -- |
| root to Raf | Yes | No | No | -- |
| root to ~Raf | Yes | No | Yes | hop |
| ~Raf to root | Yes | No | Yes | root |
| Raf to Int | Yes | No | Yes | root |
| Int to Raf | Yes | No | Yes | raf |
| ~Raf to Int | Yes | No | Yes | root |
| Int to ~Raf | Yes | No | Yes | hop |
| Raf to Raf | Yes | No | No | -- |
| Raf to ~Raf | Yes | No | Yes | hop |
| ~Raf to Raf | Yes | No | Yes | dst |
| ~Raf to ~Raf | Yes | No | Yes | hop |
In storing mode, RFC 6553 (RPI) is used to send RPL Information instanceID and rank information.
As stated in Section 16.2 of [RFC6550] a RPL-aware-leaf node does not generally issue DIO messages; a leaf node accepts DIO messages from upstream. (When the inconsistency in routing occurs, a leaf node will generate a DIO with an infinite rank, to fix it). It may issue DAO and DIS messages though it generally ignores DAO and DIS messages.
In storing mode, RFC 6553 (RPI) is used to send RPL Information instanceID and rank information.
In this case the flow comprises:
RPL-aware-leaf (6LN) --> 6LR --> 6LR,... --> root (6LBR)
As it was mentioned In this document 6LRs, 6LBR are always full-fledge RPL routers.
The 6LN inserts the RPI header, and sends the packet to 6LR which decrements the rank in RPI and sends the packet up. When the packet arrives at 6LBR, the RPI is removed and the packet is processed.
No IP-in-IP header is required.
The RPI header can be removed by the 6LBR because the packet is addressed to the 6LBR. The 6LN must know that it is communicating with the 6LBR to make use of this scenario. The 6LN can know the address of the 6LBR because it knows the address of the root via the DODAGID in the DIO messages.
| Header | 6LN | 6LR | 6LBR |
|---|---|---|---|
| Inserted headers | RPI | -- | -- |
| Removed headers | -- | -- | RPI |
| Re-added headers | -- | -- | -- |
| Modified headers | -- | RPI | -- |
| Untouched headers | -- | -- | -- |
In this case the flow comprises:
root (6LBR)--> 6LR --> RPL-aware-leaf (6LN)
In this case the 6LBR inserts RPI header and sends the packet down, the 6LR is going to increment the rank in RPI (examines instanceID for multiple tables), the packet is processed in 6LN and RPI removed.
No IP-in-IP header is required.
| Header | 6LBR | 6LR | 6LN |
|---|---|---|---|
| Inserted headers | RPI | -- | -- |
| Removed headers | -- | -- | RPI |
| Re-added headers | -- | -- | -- |
| Modified headers | -- | RPI | -- |
| Untouched headers | -- | -- | -- |
In this case the flow comprises:
root (6LBR)--> 6LR --> not-RPL-aware-leaf (6LN)
As the RPI extension can be ignored by the not-RPL-aware leaf, this situation is identical to the previous scenario.
| Header | 6LBR | 6LR | IPv6 |
|---|---|---|---|
| Inserted headers | -- | -- | -- |
| Removed headers | -- | -- | -- |
| Re-added headers | -- | -- | -- |
| Modified headers | -- | -- | -- |
| Untouched headers | -- | -- | -- |
In this case the flow comprises:
not-RPL-aware-leaf (6LN) --> 6LR --> root (6LBR)
When the packet arrives from IPv6 node to 6LR, the 6LR will insert an RPI header, encapsuladed in a IPv6-in-IPv6 header. The IPv6-in-IPv6 header can be addressed to the next hop, or to the root. The root removes the header and processes the packet.
| Header | IPv6 | 6LR | 6LBR |
|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | -- |
| Removed headers | -- | -- | IP-in-IP(RPI) |
| Re-added headers | -- | -- | -- |
| Modified headers | -- | -- | -- |
| Untouched headers | -- | -- | -- |
RPL information from RFC 6553 MAY go out to Internet as it will be ignored by nodes which have not been configured to be RPI aware.
In this case the flow comprises:
RPL-aware-leaf (6LN) --> 6LR --> root (6LBR) --> Internet
No IP-in-IP header is required.
| Header | 6LN | 6LR | 6LBR | Internet |
|---|---|---|---|---|
| Inserted headers | RPI | -- | -- | -- |
| Removed headers | -- | -- | -- | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | RPI | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
Internet --> root (6LBR) --> 6LR --> RPL-aware-leaf (6LN)
When the packet arrives from Internet to 6LBR the RPI header is added in a outer IPv6-in-IPv6 header and sent to 6LR, which modifies the rank in the RPI. When the packet arrives at 6LN the RPI header is removed and the packet processed.
| Header | Internet | 6LBR | 6LR | 6LN |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | -- | -- |
| Removed headers | -- | -- | -- | IP-in-IP(RPI) |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | RPI | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
not-RPL-aware-leaf (6LN) --> 6LR --> root (6LBR) --> Internet
The 6LR node will add an IP-in-IP(RPI) header addressed either to the root, or hop-by-hop such that the root can remove the RPI header before passing upwards.
The originating node will ideally leave the IPv6 flow label as zero so that it can be better compressed through the LLN, and the 6LBR will set the flow label to a non-zero value when sending to the Internet.
| Header | 6LN | 6LR | 6LBR | Internet |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | -- | -- |
| Removed headers | -- | -- | IP-in-IP(RPI) | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
Internet --> root (6LBR) --> 6LR --> not-RPL-aware-leaf (6LN)
The 6LBR will have to add an RPI header within an IP-in-IP header. The IP-in-IP can be addressed to the not-RPL-aware-leaf, leaving the RPI inside.
The 6LBR MAY set the flow label on the inner IP-in-IP header to zero in order to aid in compression, as the packet will not emerge again from the LLN.
| Header | Internet | 6LBR | 6LR | IPv6 |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | -- | -- |
| Removed headers | -- | -- | IP-in-IP(RPI) | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In [RFC6550] RPL allows a simple one-hop optimization for both storing and non-storing networks. A node may send a packet destined to a one-hop neighbor directly to that node. Section 9 in [RFC6550].
In this case the flow comprises:
6LN --> 6LR --> common parent (6LR) --> 6LR --> 6LN
This case is assumed in the same RPL Domain. In the common parent, the direction of RPI is changed (from increasing to decreasing the rank).
While the 6LR nodes will update the RPI, no node needs to add or remove the RPI, so no IP-in-IP headers are necessary. This may be done regardless of where the destination is, as the included RPI will be ignored by the receiver.
| Header | 6LN src | 6LR | 6LR (common parent) | 6LR | 6LN dst |
|---|---|---|---|---|---|
| Inserted headers | RPI | -- | -- | -- | -- |
| Removed headers | -- | -- | -- | -- | RPI |
| Re-added headers | -- | -- | -- | -- | -- |
| Modified headers | -- | RPI (decreasing rank) | RPI (increasing rank) | -- | -- |
| Untouched headers | -- | -- | -- | -- | -- |
In this case the flow comprises:
6LN --> 6LR --> common parent (6LR) --> 6LR --> not-RPL-aware 6LN
This situation is identical to the situation Section 5.9
In this case the flow comprises:
not-RPL-aware 6LN --> 6LR --> common parent (6LR) --> 6LR --> 6LN
The 6LR receives the packet from the the IPv6 node and inserts and the RPI header encapsulated in IPv6-in-IPv6 header. The IP-in-IP header is addressed to the destinion 6LN.
| Header | IPv6 | 6LR | common parent (6LR) | 6LR | 6LN |
|---|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | -- | -- | -- |
| Removed headers | -- | -- | -- | -- | IP-in-IP(RPI) |
| Re-added headers | -- | -- | -- | -- | -- |
| Modified headers | -- | -- | IP-in-IP(RPI) | IP-in-IP(RPI) | -- |
| Untouched headers | -- | -- | -- | -- | -- |
In this case the flow comprises:
not-RPL-aware 6LN (IPv6 node)--> 6LR --> root (6LBR) --> 6LR --> not-RPL-aware 6LN (IPv6 node)
This flow is identical to Section 5.11
| Use Case | RPI | RH3 | IP-in-IP | IP-in-IP dst |
|---|---|---|---|---|
| Raf to root | Yes | No | No | -- |
| root to Raf | Yes | Yes | No | -- |
| root to ~Raf | No | Yes | Yes | 6LR |
| ~Raf to root | Yes | No | Yes | root |
| Raf to Int | Yes | No | Yes | root |
| Int to Raf | opt | Yes | Yes | dst |
| ~Raf to Int | Yes | No | Yes | root |
| Int to ~Raf | opt | Yes | Yes | 6LR |
| Raf to Raf | Yes | Yes | Yes | root/dst |
| Raf to ~Raf | Yes | Yes | Yes | root/6LR |
| ~Raf to Raf | Yes | Yes | Yes | root/6LN |
| ~Raf to ~Raf | Yes | Yes | Yes | root/6LR |
In non-storing mode the leaf node uses default routing to send traffic to the root. The RPI header must be included to avoid/detect loops.
RPL-aware-leaf (6LN) --> 6LR --> root (6LBR)
This situation is the same case as storing mode.
| Header | 6LN | 6LR | 6LBR |
|---|---|---|---|
| Inserted headers | RPI | -- | -- |
| Removed headers | -- | -- | RPI |
| Re-added headers | -- | RPI | -- |
| Modified headers | -- | -- | -- |
| Untouched headers | -- | -- | -- |
In this case the flow comprises:
root (6LBR)--> 6LR --> RPL-aware-leaf (6LN)
The 6LBR will insert an RH3, and may optionally insert an RPI header. No IP-in-IP header is necessary as the traffic originates with an RPL aware node, the 6LBR. The destination is known to 6LBR because, the root knows the whole topology in non-storing mode.
| Header | 6LBR | 6LR | 6LN |
|---|---|---|---|
| Inserted headers | (opt: RPI), RH3 | -- | -- |
| Removed headers | -- | -- | RH3,RPI |
| Re-added headers | -- | -- | -- |
| Modified headers | -- | RH3 | -- |
| Untouched headers | -- | -- | -- |
In this case the flow comprises:
root (6LBR)--> 6LR --> not-RPL-aware-leaf (IPv6 node)
In 6LBR the RH3 is added, modified in each intermediate 6LR and it is fully consumed in the last 6LR, but left there. If the RPI is left present, the IPv6 node which does not understand it will drop it, therefore the RPI should be removed before reaching the IPv6-only node. To permit removal, an IP-in-IP header (hop-by-hop) or addressed to the last 6LR is necessary. Due the complete knowledge of the topology at the root, the 6LBR is able to address the IP-in-IP header to the last 6LR.
Omitting the RPI entirely is therefore a better solution, as no IP-in-IP header is necessary.
| Header | 6LBR | 6LR | IPv6 |
|---|---|---|---|
| Inserted headers | RH3 | -- | -- |
| Removed headers | -- | -- | -- |
| Re-added headers | -- | -- | -- |
| Modified headers | -- | RH3 | -- |
| Untouched headers | -- | -- | -- |
In this case the flow comprises:
IPv6-node --> 6LR1 --> 6LR2 --> root (6LBR)
In this case the RPI is added by the first 6LR, encapsulated in an IP-in-IP header, and is not modified in the followings 6LRs. The RPI and entire packet is consumed by the root.
| Header | IPv6 | 6LR1 | 6LR2 | 6LBR |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | -- | -- |
| Removed headers | -- | -- | -- | IP-in-IP(RPI) |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | IP-in-IP(RPI) | -- | -- |
In this case the flow comprises:
RPL-aware-leaf (6LN) --> 6LR --> root (6LBR) --> Internet
This case requires that the RPI be added, but removed by the 6LBR. The 6LN must therefore add the RPI inside an IP-in-IP header, addressed to the root. This case is identical to storing-mode case.
The IPv6 flow label should be set to zero to aid in compression, and the 6LBR will set it to a non-zero value when sending towards the Internet.
| Header | 6LN | 6LR | 6LBR | Internet |
|---|---|---|---|---|
| Inserted headers | IP-in-IP(RPI) | -- | -- | -- |
| Removed headers | -- | -- | IP-in-IP(RPI) | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | RPI | -- | -- |
In this case the flow comprises:
Internet --> root (6LBR) --> 6LR --> RPL-aware-leaf (6LN)
The 6LBR must add an RH3 header. As the 6LBR will know the path and address of the target not, it can address the IP-in-IP header to that node. The 6LBR will zero the flow label upon entry in order to aid compression.
The RPI may be added or not.
| Header | Internet | 6LBR | 6LR | 6LN |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RH3,opt:RPI) | -- | -- |
| Removed headers | -- | -- | IP-in-IP(RH3) | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | IP-in-IP(RH3) | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
not-RPL-aware-leaf (6LN) --> 6LR --> root (6LBR) --> Internet
In this case the flow label is recommended to be zero in the IPv6 node. As RPL headers are added in the IPv6 node, the first 6LN will add an RPI header inside a new IP-in-IP header. The IP-in-IP header will be addressed to the root. This case is identical to the storing-mode case (Section 5.7).
| Header | IPv6 | 6LR | 6LBR | Internet |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | -- | -- |
| Removed headers | -- | -- | IP-in-IP(RPI) | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
Internet --> root (6LBR) --> 6LR --> not-RPL-aware-leaf (IPv6 node)
The 6LBR must add an RH3 header inside an IP-in-IP header. The 6LBR will know the path, and will recognize that the final node is not an RPL capable node as it will have received the connectivity DAO from the nearest 6LR. The 6LBR can therefore make the IP-in-IP header destination be the last 6LR. The 6LBR will set to zero the flow label upon entry in order to aid compression.
| Header | Internet | 6LBR | 6LR | IPv6 |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RH3,opt:RPI) | -- | -- |
| Removed headers | -- | -- | IP-in-IP(RH3, RPI) | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
6LN --> 6LR --> root (6LBR) --> 6LR --> 6LN
This case involves only nodes in same RPL Domain. The originating node will add an RPI header to the original packet, and send the packet upwards.
The originating node could put the RPI into an IP-in-IP header addressed to the root, so that the 6LBR can remove that header.
The 6LBR will need to insert an RH3 header, which requires that it add an IP-in-IP header. It may be able to remove the RPI if it was contained in an IP-in-IP header addressed to it. Otherwise, there may be an RPI header buried inside the inner IP header, which should get ignored.
Networks that use the RPL P2P extension [RFC6997] are essentially non-storing DODAGs and fall into this scenario.
| Header | 6LN src | 6LBR | 6LR | 6LN dst |
|---|---|---|---|---|
| Inserted headers | IP-in-IP(RPI) | IP-in-IP(RH3 to 6LN,RPI) | -- | -- |
| Removed headers | -- | -- | -- | IP-in-IP(RH3,RPI) |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
6LN --> 6LR --> root (6LBR) --> 6LR --> not-RPL-aware (IPv6 node)
As in the previous case, the 6LN will insert an RPI header which MUST be in an IP-in-IP header addressed to the root so that the 6LBR can remove this RPI. The 6LBR will then insert an RH3 inside a new IP-in-IP header addressed to the 6LN above the destination node.
| Header | 6LN | 6LBR | 6LR | IPv6 |
|---|---|---|---|---|
| Inserted headers | IP-in-IP(RPI) | IP-in-IP(RH3, opt RPI) | -- | -- |
| Removed headers | -- | IP-in-IP(RPI) | IP-in-IP(RH3, opt RPI) | -- |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
not-RPL-aware 6LN --> 6LR --> root (6LBR) --> 6LR --> 6LN
This scenario is mostly identical to the previous one. The RPI is added by the first 6LR inside an IP-in-IP header addressed to the root. The 6LBR will remove this RPI, and add it's own IP-in-IP header containing an RH3 header.
| Header | IPv6 | 6LR | 6LBR | 6LN |
|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | IP-in-IP(RH3) | -- |
| Removed headers | -- | IP-in-IP(RPI) | -- | IP-in-IP(RH3) |
| Re-added headers | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- |
In this case the flow comprises:
not-RPL-aware 6LN --> 6LR --> root (6LBR) --> 6LR --> not-RPL-aware (IPv6 node)
This scenario is the combination of the previous two cases.
| Header | IPv6 | 6LR | 6LBR | 6LR | IPv6 |
|---|---|---|---|---|---|
| Inserted headers | -- | IP-in-IP(RPI) | IP-in-IP(RH3) | -- | -- |
| Removed headers | -- | -- | IP-in-IP(RPI) | IP-in-IP(RH3, opt RPI) | -- |
| Re-added headers | -- | -- | -- | -- | -- |
| Modified headers | -- | -- | -- | -- | -- |
| Untouched headers | -- | -- | -- | -- | -- |
In the completely general storing case, which includes not-RPL aware leaf nodes, it is not possible for a sending node to know if the destination is RPL aware, and therefore it must always use hop-by-hop IP-in-IP encapsulation, and it can never omit the IP-in-IP encapsulation. See table Table 1
The simplest fully general approach for storing mode is to always put in hop-by-hop IP-in-IP headers. [I-D.ietf-roll-routing-dispatch] shows that this hop-by-hop IP-in-IP header can be compressed down to {TBD} bytes.
There are potential significant advantages to having a single code path that always processes IP-in-IP headers with no options.
If all RPL aware nodes can be told/configured that there are no non-RPL aware leaf nodes, then the only case where an IP-in-IP header is needed is when communicating outside the LLN. The 6LBR knows well when the communication is from the outside, and the 6LN can tell by comparing the destination address to the prefix provided in the PIO. If it is known that there are no communications outside the RPL domain (noting that the RPL domain may well extend to outside the LLN), then RPI headers can be included in all packets, and IP-in-IP headers are *never* needed. This may be significantly advantageous in relatively closed systems such as in building or industrial automation. Again, there are advantages to having a single code path.
In order to support the above two cases with full generality, the different situations (always do IP-in-IP vs never use IP-in-IP) should be signaled in the RPL protocol itself.
In the non-storing case, dealing with non-RPL aware leaf nodes is much easier as the 6LBR (DODAG root) has complete knowledge about the connectivity of all DODAG nodes, and all traffic flows through the root node.
The 6LBR can recognize non-RPL aware leaf nodes because it will receive a DAO about that node from the 6LN immediately above that node. This means that the non-storing mode case can avoid ever using hop-by-hop IP-in-IP headers.
It is unclear what it would mean for an RH3 header to be present in a hop-by-hop IP-in-IP header. The receiving node ought to consume the IP-in-IP header, and therefore consume the RH3 as well, and then attempt to send the packet again. But intermediate 6LN nodes would not know how to forward the packet (because they do not save the sate), so the RH3 would need to be retained. This is a new kind of IPv6 packet processing. Therefore it may be that on the outbound leg of non-storing RPL networks, that hop-by-hop IP-in-IP header can NOT be used.
[I-D.ietf-roll-routing-dispatch] shows how the destination=root, and destination=6LN IP-in-IP header can be compressed down to {TBD} bytes.
Unlike in the storing mode case, there is no need for all nodes to know about the existence of non-RPL aware nodes. Only the 6LBR needs to change when there are non-RPL aware nodes. Further, in the non-storing case, the 6LBR is informed by the DAOs when there are non-RPL aware nodes.
The [I-D.ietf-roll-routing-dispatch] proposes a compression method for RPI, RH3 and IPv6-in-IPv6.
In Storing Mode, for the examples of Flow from RPL-aware-leaf to non-RPL-aware-leaf and non-RPL-aware-leaf to non-RPL-aware-leaf comprise an IP-in-IP and RPI compression headers. The type of this case is critical since IP-in-IP is encapsulating a RPI header.
+--+-----+---+--------------+-----------+-------------+-------------+ |1 | 0|0 |TSE| 6LoRH Type 6 | Hop Limit | RPI - 6LoRH | LOWPAN IPHC | +--+-----+---+--------------+-----------+-------------+-------------+
Figure 3: Critical IP-in-IP (RPI).
There are no IANA considerations related to this document.
The security considerations covering of [RFC6553] and [RFC6554] apply when the packets get into RPL Domain.
This work is partially funded by the FP7 Marie Curie Initial Training Network (ITN) METRICS project (grant agreement No. 607728).
The authors would like to acknowledge the review, feedback, and comments of Robert Cragie, Simon Duquennoy, Cenk Gündogan, Peter van der Stok, Xavier Vilajosana and Thomas Watteyne.