SPRING Working Group Z. Ali
Internet Draft C. Filsfils
Intended status: Standards Track N. Kumar
Expires: June 23, 2018 C. Pignataro
F. Iqbal
Cisco Systems, Inc.
J. Leddy
Comcast
S. Matsushima
SoftBank
R. Raszuk
Bloomberg LP
B. Peirens
Proximus
G. Naik
Drexel University
December 23, 2017
Operations, Administration, and Maintenance (OAM) in Segment Routing
Networks with IPv6 Dataplane (SRv6)
draft-spring-srv6-oam-00.txt
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
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."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/1id-abstracts.html
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html
This Internet-Draft will expire on June 23, 2018.
ali, et al. Expires June 23, 2018 [Page 1]
Internet-Draft SRv6 OAM December 2017
Copyright Notice
Copyright (c) 2017 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.
Abstract
This document outlines various use-cases for Operations,
Administration, and Maintenance (OAM) in Segment Routing with the
IPv6 data plane (SRv6) network. It also specifies solutions to
address the SRv6 OAM requirements.
Table of Contents
1. Introduction...................................................3
1.1. Terminology and Reference Topology........................3
2. Use-cases......................................................4
2.1. Connectivity Verification.................................5
2.2. Monitoring a Specific Flow................................5
2.3. Monitoring all ECMP/ UCMP Paths...........................5
2.4. Traceroute................................................6
2.5. Proof of Transit..........................................6
2.6. Anycast Server selection..................................7
2.7. Detecting Path Divergence.................................7
2.8. Fault Isolation...........................................7
2.9. Connectivity Verification from arbitrary node.............7
3. OAM Mechanisms.................................................8
3.1. Ping......................................................8
3.1.1. Classic Ping.........................................8
3.1.2. Pinging a SID Function...............................9
3.1.2.1. End-to-end ping using EDN.OTP..................10
3.1.2.2. Segment-by-segment ping using O-bit (Proof of
Transit)................................................11
3.2. Error Reporting..........................................12
3.3. Traceroute...............................................12
3.3.1. Classic Traceroute..................................13
3.3.2. Traceroute to a SID Function........................14
ali, et al. Expires June 23, 2018 [Page 2]
Internet-Draft SRv6 OAM December 2017
3.3.2.1. Hop-by-hop traceroute using END.OTP............15
3.3.2.2. Tracing SRv6 Overlay...........................16
3.4. In-situ OAM..............................................18
3.5. Seamless BFD Applicability...............................19
3.6. Connectivity Verification from arbitrary SR node.........19
4. Security Considerations.......................................20
5. IANA Considerations...........................................20
6. References....................................................20
6.1. Normative References.....................................20
6.2. Informative References...................................21
7. Acknowledgments...............................................21
1. Introduction
This document outlines various SRv6 OAM use-cases. It also describes
OAM mechanisms that can be used to address SRv6 OAM requirements.
Additional OAM use-cases and mechanisms will be added in a future
revision of the document.
1.1. Terminology and Reference Topology
This document uses the terminology defined in [I-D.draft-filsfils-
spring-srv6-network-programming]. The readers are expected to be
familiar with the same.
Throughout the document, the following simple topology is used for
illustration.
+--------------------------| N100 |------------------------+
| |
====== link1====== link3------ link5====== link9------
||N1||======||N2||======| N3 |======||N4||======| N5 |
|| ||------|| ||------| |------|| ||------| |
====== link2====== link4------ link6======link10------
| |
| ------ |
+-------| N6 |---------+
link7 | | link8
------
Reference Topology
In the reference topology:
Nodes N1, N2, and N4 are SRv6 capable nodes.
ali, et al. Expires June 23, 2018 [Page 3]
Internet-Draft SRv6 OAM December 2017
Nodes N3, N5 and N6 are classic IPv6 nodes.
Node 100 is a controller.
Node Nk has a classic IPv6 loopback address Bk::/128
Node Nk has Ak::/48 for its local SID space from which Local SIDs
are explicitly allocated.
The IPv6 address of the nth Link between node X and Y at the X side
is represented as 99:X:Y::Xn. e.g., the IPv6 address of link6 (the
2nd link) between N3 and N4 at N3 in Figure 1 is 99:3:4:32.
Similarly, the IPv6 address of link5 (the 1st link between N3 and
N4) at node 3 is 99:3:4::31.
Ak::0 is explicitly allocated as the END function at Node k.
Ak::Cij is explicitly allocated as the END.X function at node k
towards neighbor node i via jth Link between node i and node j.
e.g., A2::C31 represents END.X at N2 towards N3 via link3 (the 1st
link between N2 and N3). Similarly, A4::C52 represents the END.X at
N4 towards N5 via link10.
SRH is the abbreviation for the Segment Routing Header.
SL is the abbreviation for the Segment Left.
SID is the abbreviation for the Segment ID.
<S1, S2, S3> represents a SID list where S1 is the first SID and S3
is the last SID. (S3, S2, S1; SL) represents the same SID list but
encoded in the SRH format where the rightmost SID (S1) in the SRH is
the first SID and the leftmost SID (S3) in the SRH is the last SID.
ECMP is the abbreviation for the Equal Cost Multi-Path.
UCMP is the abbreviation for the Unequal Cost Multi-Path.
2. Use-cases
This section outlines some for the basic OAM use-cases in an SRv6
network. Additional use-cases will be added in a future revision of
the document.
ali, et al. Expires June 23, 2018 [Page 4]
Internet-Draft SRv6 OAM December 2017
2.1. Connectivity Verification
One of the basic OAM use-cases for any network is the capability to
perform path monitoring between different end points over any
possible shortest path without any path preference. Such essential
path monitoring helps to monitor the path availability and the
liveliness of the remote end point.
The shortest path monitoring can be done continuously or can be
triggered on demand basis using an external event like a script or a
CLI trigger. It may be required to perform the connectivity
verification in the order of milliseconds, or at a slower pace.
In the reference topology in Figure 1, N1 can send OAM probe packet
destined to loopback address of N5 (B5::) to monitor the path
liveliness between N1 and N5. N1 optionally may include any relevant
segment list in SRH. N1 is not concerned about which route is taken
by the probe between N1 and N5 as long as N1 receives the response
back from N5. All transit nodes treat the probe packet as like other
data packet and forward it based on the Destination Address (DA). N5
looks into the payload of probe packet and respond back to the
source address of the probe packet (N1).
2.2. Monitoring a Specific Flow
The network OAM needs to have the ability to monitor a particular
path from the available ECMP paths. For example, in the reference
topology in figure 1, there are many ECMP paths between N1 and N5.
However, the service provider may like to monitor a flow that
follows [N1]-<link1>-[N2]-<link7>-[N6]-<link8>-[N4]-<link9>-[N5].
The flow monitoring can be done continuously or can be triggered on
demand basis. It may be required to perform the connectivity
verification in the order of milliseconds, or at a slower pace.
2.3. Monitoring all ECMP/ UCMP Paths
In any network, it is common to see multiple ECMP paths between end
points that are used for load balancing or redundancy. While
monitoring, the shortest path helps to monitor the path and
liveliness of remote node, it may not be sufficient to detect any
failure in one of the ECMP paths. In our reference topology in
figure 1, N6 has 2 ECMP paths to reach N5 as below:
N6--<link8>--N4--<link9>--N5
N6--<link8>--N4--<link10>--N5
ali, et al. Expires June 23, 2018 [Page 5]
Internet-Draft SRv6 OAM December 2017
If the probe packet from N6 to N5 uses link10, it may not detect any
failure on link9. It is critical and beneficial to discover and
monitor all ECMP/ UCMP paths. Monitoring of all ECMP/ UCMP paths can
be done by probing the candidate paths from end-to-end or by each
node by monitoring its data plane.
2.4. Traceroute
It is essential to trace the path between different end points for
troubleshooting and fault localization purpose. In the SRv6 network,
depending on the forwarding instruction encoded in SRH, a packet may
traverse over zero or more SRv6 transit nodes which in turn are
connected through transit IPv6 nodes. For example, the best effort
traffic may traverse the shortest path between Ingress and egress
nodes while an SLA constrained traffic may follow a specific path
that involves one or more transit SRv6 nodes.
In either of these cases, traceroute functionality allows an
operator to discover the set of SRv6 and/or IPv6 nodes along the
path between different end points. Multipath being inevitable in any
network, it is also essential to identify the exact path (among the
available equal cost multi paths) that a particular flow or packet
is traversing.
2.5. Proof of Transit
Various scenarios require the packet to be steered over a particular
links or nodes. For example:
- Voice traffic in a SLA constrained network needs to traverse a
low latency path between endpoints which may not be the shortest
path, i.e. the voice traffic needs to be traffic engineered and
steered over the specified segment list that satisfies the SLA
constraint.
- In a service chaining environment, the traffic may need to
traverse over an ordered list of service functions.
In these scenarios, the SRH contains the list of SID functions that
the packet should execute before reaching the destination. It is
possible, due to an error, that the packet may reach the destination
without visiting all the segments in the segment list. It is,
therefore, important to have the ability to verify that all the
function SIDs have been executed correctly before the packet is
delivered to the destination. It is also important to ensure that
ali, et al. Expires June 23, 2018 [Page 6]
Internet-Draft SRv6 OAM December 2017
the order of execution of the SID function has been consistent with
the SRH contents.
2.6. Anycast Server selection
For application redundancy and load sharing purpose, it is prevalent
to see anycast deployment where the service address will assign to
different application servers spanned across the network. While
traditionally this type of deployment model was used to terminate
the client session to the nearest server, the recent capability of
collecting network and application telemetry along with the traffic
steering characteristics of SRv6 allows an operator to leverage the
knowledge to choose the right server and path based on not just the
shortest path, but also based on other performance metrics.
It is therefore essential to have the ability to monitor the anycast
server performance and detect any deviation and take corrective
actions.
2.7. Detecting Path Divergence
Path divergence occurs when network traffic diverges from the
expected path that packet was supposed to take. Path divergence may
result in congestion, delay, or breakage of strict SLAs promised to
customers. It is, therefore, important to exercise mechanisms that
can detect path divergence in the SRv6 network.
2.8. Fault Isolation
In the cases where a monitoring technique discovers an issue, it is
required to have the ability to pinpoint the failure location. The
fault isolation mechanisms are required to help service providers
troubleshoot failure in an SRv6 network.
2.9. Connectivity Verification from arbitrary node
In the recent past, network operators are interested in performing
network operations, administration, and maintenance configuration in
a centralized manner. In this use-case, one of the requirements is
to implement OAM functionality like connectivity verification
between different SRv6 end points in a centralized manner by
triggering it from any arbitrary node. The other requirement in this
use-case is to perform the connectivity verification between end
points without any control plane intervention at the monitored or
other transit nodes.
ali, et al. Expires June 23, 2018 [Page 7]
Internet-Draft SRv6 OAM December 2017
Additional OAM use-cases will be included in a future revision of
the document.
3. OAM Mechanisms
This section describes how ping and traceroute mechanisms can be
used in an SRv6 network. Additional OAM mechanisms will be added in
a future revision of the document.
3.1. Ping
[RFC4443] describes Internet Control Message Protocol for IPv6
(ICMPv6) that is used by IPv6 devices for network diagnostic and
error reporting purposes. As Segment Routing with IPv6 data plane
(SRv6) simply adds a new type of Routing Extension Header, existing
ICMPv6 mechanisms can be used in an SRv6 network. This section
describes the applicability of ICMPv6 in the SRv6 network and how
the existing ICMPv6 mechanisms can be used for providing OAM
functionality to address many use-cases outlined in Section 2.
Throughout this document, unless otherwise specified, the acronym
ICMPv6 refers to multi-part ICMPv6 messages [RFC4884]. The document
does not propose any changes to the standard ICMPv6 [RFC4443],
[RFC4884] or standard ICMPv4 [RFC792].
There is no hardware or software change required for ping operation
at the classic IPv6 nodes in an SRv6 network. That includes the
classic IPv6 node with ingress, egress or transit roles.
Furthermore, no protocol changes are required to the standard ICMPv6
[RFC4443], [RFC4884] or standard ICMPv4 [RFC792]. In other words,
existing ICMP ping mechanisms work seamlessly in the SRv6 networks.
The following subsections outline some use cases of the ICMP ping in
the SRv6 networks.
3.1.1. Classic Ping
The existing mechanism to ping a remote IP prefix, along the
shortest path, continues to work without any modification. The
initiator may be an SRv6 node or a classic IPv6 node. Similarly, the
egress or transit may be an SRv6 capable node or a classic IPv6
node.
If an SRv6 capable ingress node wants to ping an IPv6 prefix via an
arbitrary segment list <S1, S2, S3>, it needs to initiate ICMPv6
ping with an SR header containing the SID list <S1, S2, S3>. This is
illustrated using the topology in Figure 1. Assume all the links
ali, et al. Expires June 23, 2018 [Page 8]
Internet-Draft SRv6 OAM December 2017
have IGP metric 10 except both links between node2 and node3, which
have IGP metric set to 100. User issues a ping from node N1 to a
loopback of node 5, via via segment list <A2::C31, A4::C52>.
Figure 2 contains sample output for a ping request initiated at node
N1 to the loopback address of node N5 via a segment list <A2::C31,
A4::C52>.
> ping B5:: via segment-list A2::C31, A4::C52
Sending 5, 100-byte ICMP Echos to B5::, timeout is 2 seconds:
!!!!!
Success rate is 100 percent (5/5), round-trip min/avg/max = 0.625
/0.749/0.931 ms
A sample ping output at an SRv6 capable node
All transit nodes process the echo request message like any other
data packet carrying SR header and hence do not require any change.
Similarly, the egress node (IPv6 classic or SRv6 capable) does not
require any change to process the ICMPv6 echo request. For example,
in the ping example of Figure 2:
- Node N1 initiates an ICMPv6 ping packet with SRH as follows (B1::,
A2::C31)(B1::, A4::C52, A2::C31, SL=2, NH: ICMPv6)(ICMPv6 Echo
Request).
- Node N2, which is an SRv6 capable node, performs the standard SRH
processing. Specifically, it executes the END.X function (A2::C31)
on the echo request packet.
- Node N3, which is a classic IPv6 node, performs the standard IPv6
processing. Specifically, it forwards the echo request based on DA
A4::C52 in the IPv6 header.
- Node N4, which is an SRv6 capable node, performs the standard SRH
processing. Specifically, it observes the END.X function (A4::C52)
with PSP (Penultimate Segment POP) on the echo request packet and
removes the SRH and forwards the packet across link10 to N5.
- The echo request packet at N5 arrives as an IPv6 packet without a
SRH. Node N5, which is a classic IPv6 node, performs the standard
IPv6/ ICMPv6 processing on the echo request and responds,
accordingly.
3.1.2. Pinging a SID Function
The classic ping described in the previous section cannot be used to
ping a remote SID function, as explained using an example in the
following. Consider the case where the user wants to ping the remote
SID function A4::C52, via A2::C31, from node N1. Node N1 constructs
the ping packet (B1::0, A2::C31)( A4::C52, A2::C31, SL=1;
ali, et al. Expires June 23, 2018 [Page 9]
Internet-Draft SRv6 OAM December 2017
NH=ICMPv6)(ICMPv6 Echo Request). When the node N4 receives the
ICMPv6 echo request with DA set to A4::C52 and next header set to
ICMPv6, it silently drops it (see [I-D.draft-filsfils-spring-srv6-
network-programming] for details). To solve this problem, the
initiator needs to mark the ICMPv6 echo request as an OAM packet.
The OAM packets are identified either by setting the O-bit in SRH or
by inserting the END.OTP SID at an appropriate place in the SRH [I-
D.draft-filsfils-spring-srv6-network-programming].
In an SRv6 network, the user can exercise two flavors of the ping:
end-to-end ping or segment-by-segment ping, as outlined in the
following.
3.1.2.1. End-to-end ping using EDN.OTP
Consider the same example where the user wants to ping a remote SID
function A4::C52 , via A2::C31, from node N1. To force a punt of the
ICMPv6 echo request at the node N4, node N1 inserts the END.OTP SID
just before the target SID A4::C52 in the SRH. The ICMPv6 echo
request is processed at the individual nodes along the path as
follows.
- Node N1 initiates an ICMPv6 ping packet with SRH as follows
(B1::0, A2::C31)(A4::C52, A4::OTP, A2::C31; SL=2;
NH=ICMPv6)(ICMPv6 Echo Request).
- Node N2, which is an SRv6 capable node, performs the standard SRH
processing. Specifically, it executes the END.X function (A2::C31)
on the echo request packet.
- Node N3 receives the packet as follows (B1::0, A4::OTP)(A4::C52,
A4::OTP, A2::C31 ; SL=1; NH=ICMPv6)(ICMPv6 Echo Request). Node N3,
which is a classic IPv6 node, performs the standard IPv6
processing. Specifically, it forwards the echo request based on DA
A4::OTP in the IPv6 header.
- When node N4 receives the packet (B1::0, A4::OTP)(A4::C52,
A4::OTP, A2::C31 ; SL=1; NH=ICMPv6)(ICMPv6 Echo Request), it
processes the END.OTP SID, as described in the pseudocode in [I-
D.draft-filsfils-spring-srv6-network-programming]. The packet gets
punted to the ICMPv6 process for processing. The ICMPv6 process
checks if the next SID in SRH (the target SID A4::C52) is locally
programmed. If the target SID is not locally programmed, N4
responses with the ICMPv6 message (Type: "SRv6 OAM (TBA)", Code:
"SID not locally implemented (TBA)"); otherwise a success is
returned.
ali, et al. Expires June 23, 2018 [Page 10]
Internet-Draft SRv6 OAM December 2017
3.1.2.2. Segment-by-segment ping using O-bit (Proof of Transit)
Consider the same example where the user wants to ping a remote SID
function A4::C52 , via A2::C31, from node N1. However, in this ping,
the node N1 wants to get a response from each segment node in the
SRH. In other words, in the segment-by-segment ping case, the node
N1 expects a response from node N2 and node N4 for their respective
local SID function.
To force a punt of the ICMPv6 echo request at node N2 and node N4,
node N1 sets the O-bit in SRH. The ICMPv6 echo request is processed
at the individual nodes along the path as follows.
- Node N1 initiates an ICMPv6 ping packet with SRH as follows
(B1::0, A2::C31)(A4::C52, A2::C31; SL=1, Flags.O=1;
NH=ICMPv6)(ICMPv6 Echo Request).
- When node N2 receives the packet (B1::0, A2::C31)(A4::C52,
A2::C31; SL=1, Flags.O=1; NH=ICMPv6)(ICMPv6 Echo Request) packet,
it processes the O-bit in SRH, as described in the pseudocode in
[I-D.draft-filsfils-spring-srv6-network-programming]. A time-
stamped copy of the packet gets punted to the ICMPv6 process for
processing. Node N2 continues to apply the A2::C31 SID function on
the original packet and forwards it, accordingly. As
SRH.Flags.O=1, Node N2 also disables the PSP flavour, i.e., does
not remove the SRH. The ICMPv6 process at node N2 checks if its
local SID (A2::C31) is locally programmed or not and responds to
the ICMPv6 Echo Request. If the target SID is not locally
programmed, N4 responses with the ICMPv6 message (Type: "SRv6 OAM
(TBA)", Code: "SID not locally implemented (TBA)"); otherwise a
success is returned. Please note that, as mentioned in [I-D.draft-
filsfils-spring-srv6-network-programming], if node N2 does not
support the O-bit, it simply ignores it and process the local SID,
A2::C31.
- Node N3, which is a classic IPv6 node, performs the standard IPv6
processing. Specifically, it forwards the echo request based on DA
A4::C52 in the IPv6 header.
- When node N4 receives the packet (B1::0, A4::C52)(A4::C52,
A2::C31; SL=0, Flags.O=1; NH=ICMPv6)(ICMPv6 Echo Request), it
processes the O-bit in SRH, as described in the pseudocode in [I-
D.draft-filsfils-spring-srv6-network-programming]. A time-stamped
copy of the packet gets punted to the ICMPv6 process for
processing. The ICMPv6 process at node N4 checks if its local SID
(A2::C31) is locally programmed or not and responds to the ICMPv6
Echo Request. If the target SID is not locally programmed, N4
responses with the ICMPv6 message (Type: "SRv6 OAM (TBA)", Code:
"SID not locally implemented (TBA)"); otherwise a success is
returned.
ali, et al. Expires June 23, 2018 [Page 11]
Internet-Draft SRv6 OAM December 2017
Support for O-bit is part of node capability advertisement. That
enables node N1 to know which segment nodes are capable of
responding to the ICMPv6 echo request. Node N1 processes the echo
responses and presents data to the user, accordingly.
Please note that segment-by-segment ping can be used to address
proof of transit use-case discussed earlier.
3.2. Error Reporting
Any IPv6 node can use ICMPv6 control messages to report packet
processing errors to the host that originated the datagram packet.
To name a few such scenarios:
- If the router receives an undeliverable IP datagram, or
- If the router receives a packet with a Hop Limit of zero, or
- If the router receives a packet such that if the router decrements
the packet's Hop Limit it becomes zero, or
- If the router receives a packet with problem with a field in the
IPv6 header or the extension headers such that it cannot complete
processing the packet, or
- If the router cannot forward a packet because the packet is larger
than the MTU of the outgoing link.
In the scenarios listed above, the ICMPv6 response also contains the
IP header, IP extension headers and leading payload octets of the
"original datagram" to which the ICMPv6 message is a response.
Specifically, the "Destination Unreachable Message", "Time Exceeded
Message", "Packet Too Big Message" and "Parameter Problem Message"
ICMPV6 messages can contain as much of the invoking packet as
possible without the ICMPv6 packet exceeding the minimum IPv6 MTU
[RFC4443], [RFC4884]. In an SRv6 network, the copy of the invoking
packet contains the SR header. The packet originator can use this
information for diagnostic purposes. For example, traceroute can use
this information as detailed in the following.
3.3. Traceroute
There is no hardware or software change required for traceroute
operation at the classic IPv6 nodes in an SRv6 network. That
includes the classic IPv6 node with ingress, egress or transit
roles. Furthermore, no protocol changes are required to the standard
traceroute operations. In other words, existing traceroute
mechanisms work seamlessly in the SRv6 networks.
The following subsections outline some use cases of the traceroute
in the SRv6 networks.
ali, et al. Expires June 23, 2018 [Page 12]
Internet-Draft SRv6 OAM December 2017
3.3.1. Classic Traceroute
The existing mechanism to traceroute a remote IP prefix, along the
shortest path, continues to work without any modification. The
initiator may be an SRv6 node or a classic IPv6 node. Similarly, the
egress or transit may be an SRv6 node or a classic IPv6 node.
If an SRv6 capable ingress node wants to traceroute to IPv6 prefix
via an arbitrary segment list <S1, S2, S3>, it needs to initiate
traceroute probe with an SR header containing the SID list <S1, S2,
S3>. That is illustrated using the topology in Figure 1. Assume all
the links have IGP metric 10 except both links between node2 and
node3, which have IGP metric set to 100. User issues a traceroute
from node N1 to a loopback of node 5, via segment list <A2::C31,
A4::C52>. Figure 3 contains sample output for the traceroute
request.
> traceroute B5:: via segment-list A2::C31, A4::C52
Tracing the route to B5::
1 99:1:2::21 0.512 msec 0.425 msec 0.374 msec
SRH: (B5::, A4::C52, A2::C31, SL=2)
2 99:2:3::31 0.721 msec 0.810 msec 0.795 msec
SRH: (B5::, A4::C52, A2::C31, SL=1)
3 99:3:4::41 0.921 msec 0.816 msec 0.759 msec
SRH: (B5::, A4::C52, A2::C31, SL=1)
4 99:4:5::52 0.879 msec 0.916 msec 1.024 msec
A sample traceroute output at an SRv6 capable node
Please note that information for hop2 is returned by N3, which is a
classic IPv6 node. Nonetheless, the ingress node is able to display
SR header contents as the packet travels through the IPv6 classic
node. This is because the "Time Exceeded Message" ICMPv6 message can
contain as much of the invoking packet as possible without the
ICMPv6 packet exceeding the minimum IPv6 MTU [RFC4443]. The SR
header is also included in these ICMPv6 messages initiated by the
classic IPv6 transit nodes that are not running SRv6 software.
Specifically, a node generating ICMPv6 message containing a copy of
the invoking packet does not need to understand the extension
header(s) in the invoking packet.
ali, et al. Expires June 23, 2018 [Page 13]
Internet-Draft SRv6 OAM December 2017
The segment list information returned for hop1 is returned by N2,
which is an SRv6 capable node. Just like for hop2, the ingress node
is able to display SR header contents for hop1.
There is no difference in processing of the traceroute probe at an
IPv6 classic node and an SRv6 capable node. Similarly, both IPv6
classic and SRv6 capable nodes use the address of the interface on
which probe was received as the source address in the ICMPv6
response. ICMP extensions defined in [RFC5837] can be used to also
display information about the IP interface through which the
datagram would have been forwarded had it been forwardable, and the
IP next hop to which the datagram would have been forwarded, the IP
interface upon which a datagram arrived, the sub-IP component of an
IP interface upon which a datagram arrived.
The information about the IP address of the incoming interface on
which the traceroute probe was received by the reporting node is
very useful. This information can also be used to verify if SID
functions A2::C31 and A4::C52 are executed correctly by N2 and N4,
respectively. Specifically, the information displayed for hop2
contains the incoming interface address 99:2:3::31 at N3. This
matches with the expected interface bound to END.X function A2::C31
(link3). Similarly, the information displayed for hop5 contains the
incoming interface address 99:4:5::52 at N5. This matches with the
expected interface bound to the END.X function A4::C52 (link10).
3.3.2. Traceroute to a SID Function
The classic traceroute described in the previous section cannot be
used to traceroute a remote SID function, as explained using an
example in the following.
Consider the case where the user wants to traceroute the remote SID
function A4::C52, via A2::C31, from node N1. Node N1 constructs the
traceroute packet (B1::0, A2::C31, HC=1)( A4::C52, A2::C31, SL=1;
NH=UDP)(traceroute probe). Even though Hop Count of the packet is
set to 1, when the node N4 receives the traceroute probe with DA set
to A4::C52 and next header set to UDP, it silently drops it (see [I-
D.draft-filsfils-spring-srv6-network-programming] for details). To
solve this problem, the initiator needs to mark the traceroute probe
as an OAM packet.
The OAM packets are identified either by setting the O-bit in SRH or
by inserting the END.OTP SID at an appropriate place in the SRH [I-
D.draft-filsfils-spring-srv6-network-programming].
ali, et al. Expires June 23, 2018 [Page 14]
Internet-Draft SRv6 OAM December 2017
In an SRv6 network, the user can exercise two flavors of the
traceroute: hop-by-hop traceroute or overlay traceroute.
In hop-by-hop traceroute, user gets responses from all nodes
including classic IPv6 transit nodes, SRv6 capable transit nodes as
well as SRv6 capable segment endpoints. E.g., consider the example
where the user wants to traceroute to a remote SID function A4::C52
, via A2::C31, from node N1. The traceroute output will also display
information about node3, which is a transit (underlay) node.
The overlay traceroute, on the other hand, does not trace the
underlay nodes. In other words, the overlay traceroute only displays
the nodes that acts as SRv6 segments along the route. I.e., in the
example where the user wants to traceroute to a remote SID function
A4::C52 , via A2::C31, from node N1, the overlay traceroute would
only display the traceroute information from node N2 and node N2 and
will not display information from node 3.
3.3.2.1. Hop-by-hop traceroute using END.OTP
In this section, hop-by-hop traceroute to a SID function is
exemplified using UDP probes. However, the procedure is equally
applicable to other implementation of traceroute mechanism.
Consider the same example where the user wants to traceroute to a
remote SID function A4::C52 , via A2::C31, from node N1. To force a
punt of the traceroute probe only at the node N4, node N1 inserts
the END.OTP SID just before the target SID A4::C52 in the SRH. The
traceroute probe is processed at the individual nodes along the path
as follows.
- Node N1 initiates a traceroute probe packet with a monotonically
increasing value of hop count and SRH as follows (B1::0,
A2::C31)(A4::C52, A4::OTP, A2::C31; SL=2; NH=UDP)(Traceroute
probe).
- When node N2 receives the packet with hop-count = 1, it processes
the hop count expiry. Specifically, the node N2 responses with the
ICMPv6 message (Type: "Time Exceeded", Code: "Time to Live
exceeded in Transit").
- When Node N2 receives the packet with hop-count > 1, it performs
the standard SRH processing. Specifically, it executes the END.X
function (A2::C31) on the traceroute probe.
- When node N3, which is a classic IPv6 node, receives the packet
(B1::0, A4::OTP)(A4::C52, A4::OTP, A2::C31 ; HC=1, SL=1;
NH=UDP)(Traceroute probe) with hop-count = 1, it processes the hop
count expiry. Specifically, the node N3 responses with the ICMPv6
ali, et al. Expires June 23, 2018 [Page 15]
Internet-Draft SRv6 OAM December 2017
message (Type: "Time Exceeded", Code: "Time to Live exceeded in
Transit").
- When node N3, which is a classic IPv6 node, receives the packet
with hop-count > 1, it performs the standard IPv6 processing.
Specifically, it forwards the traceroute probe based on DA A4::OTP
in the IPv6 header.
- When node N4 receives the packet (B1::0, A4::OTP)(A4::C52,
A4::OTP, A2::C31 ; SL=1; HC=1, NH=UDP)(Traceroute probe), it
processes the END.OTP SID, as described in the pseudocode in [I-
D.draft-filsfils-spring-srv6-network-programming]. The packet gets
punted to the traceroute process for processing. The traceroute
process checks if the next SID in SRH (the target SID A4::C52) is
locally programmed. If the target SID A4::C52 is locally
programmed, node N4 responses with the ICMPv6 message (Type:
Destination unreachable, Code: Port Unreachable). If the target
SID A4::C52 is not a local SID, node N4 silently drops the
traceroute probe.
Figure 4 displays a sample traceroute output for this example.
> traceroute srv6 A4::C52 via segment-list A2::C31
Tracing the route to SID function A4::C52
1 99:1:2::21 0.512 msec 0.425 msec 0.374 msec
SRH: (A4::C52, A4::OTP, A2::C31; SL=2)
2 99:2:3::31 0.721 msec 0.810 msec 0.795 msec
SRH: (A4::C52, A4::OTP, A2::C31; SL=1)
3 99:3:4::41 0.921 msec 0.816 msec 0.759 msec
SRH: (A4::C52, A4::OTP, A2::C31; SL=1)
A sample output for hop-by-hop traceroute to a SID function
3.3.2.2. Tracing SRv6 Overlay
The overlay traceroute does not trace the underlay nodes, i.e., only
displays the nodes that acts as SRv6 segments along the path. This
is achieved by setting the SRH.Flags.O bit.
In this section, overlay traceroute to a SID function is exemplified
using UDP probes. However, the procedure is equally applicable to
other implementation of traceroute mechanism.
ali, et al. Expires June 23, 2018 [Page 16]
Internet-Draft SRv6 OAM December 2017
Consider the same example where the user wants to traceroute to a
remote SID function A4::C52 , via A2::C31, from node N1.
- Node N1 initiates a traceroute probe with SRH as follows (B1::0,
A2::C31)(A4::C52, A2::C31; HC=64, SL=1, Flags.O=1;
NH=UDP)(Traceroute Probe). Please note that the hop-count is set
to 64 to skip the underlay nodes from tracing. The O-bit in SRH is
set to make the overlay nodes (nodes processing the SRH) respond.
- When node N2 receives the packet (B1::0, A2::C31)(A4::C52,
A2::C31; SL=1, HC=64, Flags.O=1; NH=UDP)(Traceroute Probe), it
processes the O-bit in SRH, as described in the pseudocode in [I-
D.draft-filsfils-spring-srv6-network-programming]. A time-stamped
copy of the packet gets punted to the traceroute process for
processing. Node N2 continues to apply the A2::C31 SID function on
the original packet and forwards it, accordingly. As
SRH.Flags.O=1, Node N2 also disables the PSP flavor, i.e., does
not remove the SRH. The traceroute process at node N2 checks if
its local SID (A2::C31) is locally programmed. If the SID is not
locally programmed, it silently drops the packet. Otherwise, it
performs the egress check by looking at the SL value in SRH. As SL
is not equal to zero (i.e., it's not egress node), node N2
responses with the ICMPv6 message (Type: "SRv6 OAM (TBA)", Code:
"O-bit punt at Transit (TBA)"). Please note that, as mentioned in
[I-D.draft-filsfils-spring-srv6-network-programming], if node N2
does not support the O-bit, it simply ignores it and processes the
local SID, A2::C31.
- When node N3 receives the packet (B1::0, A4::C52)(A4::C52,
A2::C31; SL=0, HC=63, Flags.O=1; NH=UDP)(Traceroute Probe),
performs the standard IPv6 processing. Specifically, it forwards
the traceroute probe based on DA A4::C52 in the IPv6 header.
Please note that there is no hop-count expiration at the transit
nodes.
- When node N4 receives the packet (B1::0, A4::C52)(A4::C52,
A2::C31; SL=0, HC=62, Flags.O=1; NH=UDP)(Traceroute Probe), it
processes the O-bit in SRH, as described in the pseudocode in [I-
D.draft-filsfils-spring-srv6-network-programming]. A time-stamped
copy of the packet gets punted to the traceroute process for
processing. The traceroute process at node N4 checks if its local
SID (A2::C31) is locally programmed. If the SID is not locally
programmed, it silently drops the packet. Otherwise, it performs
the egress check by looking at the SL value in SRH. As SL is equal
to zero (i.e., N4 is the egress node), node N4 tries to consume
the UDP probe. As UDP probe is set to access an invalid port, the
node N4 responses with the ICMPv6 message (Type: Destination
unreachable, Code: Port Unreachable).
ali, et al. Expires June 23, 2018 [Page 17]
Internet-Draft SRv6 OAM December 2017
Figure 5 displays a sample overlay traceroute output for this
example. Please note that the underlay node N3 does not appear in
the output.
> traceroute srv6 A4::C52 via segment-list A2::C31
Tracing the route to SID function A4::C52
1 99:1:2::21 0.512 msec 0.425 msec 0.374 msec
SRH: (A4::C52, A4::OTP, A2::C31; SL=2)
2 99:3:4::41 0.921 msec 0.816 msec 0.759 msec
SRH: (A4::C52, A4::OTP, A2::C31; SL=1)
A sample output for overlay traceroute to a SID function
3.4. In-situ OAM
[I-D.draft-brockners-inband-oam-requirements] describes motivation
and requirements for In-situ OAM (iOAM). iOAM records operational
and telemetry information in the data packet while the packet
traverses the network of telemetry domain. iOAM complements out-of-
band probe based OAM mechanisms such ICMP ping and traceroute by
directly encoding tracing and the other kind of telemetry
information to the regular data traffic.
[I-D.brockners-inband-oam-transport] describes transport mechanisms
for iOAM data including IPv6 and Segment Routing traffic.
furthermore, [I-D.brockners-inband-oam-data] defines information
encoding for iOAM data.
One of the application of iOAM is to perform inband traceroute. In
SRv6 network, iOAM traceroute feature can be used to trace the order
set of segment ID executed by SRv6 nodes for packet forwarding along
the packet path. This is achieved by recording the node details that
the packet traversed in the packet header itself.
Another important application of iOAM is to perform delay
measurement in anycast server scenarios. Anycast server deployment
is commonly seen for redundancy and load balancing purpose. In SRv6
network, iOAM can be used to collect the timestamp from different
anycats servers to measure the delay induced by each server within
the anycast cluster that helps to provide SLA constrainted services.
ali, et al. Expires June 23, 2018 [Page 18]
Internet-Draft SRv6 OAM December 2017
One of the other applications of iOAM is to provide the Proof of
Transit (POT). Among other features of iOAM, SRv6 networks can use
the POT feature of iOAM to verify that all the function SIDs in SRH
have been executed before the packet is delivered to the
destination. It can also ensure that the order of execution of the
SID function has been consistent with the SRH contents.
More details on various applications of iOAM in SRv6 networks will
be included in future versions of this document.
3.5. Seamless BFD Applicability
[RFC7880] defines Seamless BFD (S-BFD) architecture that simplifies
BFD mechanism and enables it to perform path monitoring in a
controlled and scalable manner. [RFC7881] describes the procedure to
perform continuity check using S-BFD in different environments
including IPv6 networks. Section 5.1 of [RFC7881] explains the
SBFDInitiator specification and procedure to initiate S-BFD control
packet in IP and MPLS network. The specification described for IP-
routed S-BFD control packet is also directly applicable to the SRv6
network.
S-BFD has a fast bootstrapping capability. Furthermore, in S-BFD,
only the ingress is required to keep BFD states; the egress and
transit node does not have any knowledge of the BFD session. These
attributes of S-BFD make it an excellent candidate for rapid failure
detection in the SRv6 network. More details on various S-BFD usage
on the SRv6 network will be included in a future version.
3.6. Connectivity Verification from arbitrary SR node
In the recent past, network operators are interested in performing
network operations, administration, and maintenance configuration in
a centralized manner. Various data models like YANG are available to
collect data from the network and manage it from a centralized
entity.
SR technology enables a centralized OAM entity to perform path
monitoring from centralized OAM entity without control plane
intervention on monitored nodes. [I.D-draft-ietf-spring-oam-usecase]
describes such a centralized OAM mechanism. Specifically, the draft
describes a procedure that can be used to perform path continuity
check between any nodes within an SR domain from a centralized
monitoring system, with minimal or no control plane intervene on the
nodes. However, the draft focuses on SR networks with MPLS data
plane. The same concept applies to the SRv6 networks. This document
ali, et al. Expires June 23, 2018 [Page 19]
Internet-Draft SRv6 OAM December 2017
describes how the concept can be used to perform path monitoring in
an SRv6 network.
In the above reference topology, N100 is the centralized monitoring
system implementing an END function A100::. In order to verify a
segment list <A2::C31, A4::C52>, N100 generates a probe packet with
SRH set to (A100::, A4::C52, A2::C31, SL=2). The controller routes
the probe packet towards the first segment, which is A2::C31. N2
performs the standard SRH processing and forward it over link3 with
the DA of IPv6 packet set to A4::C52. N4 also performs the normal
SRH processing and forward it over link10 with the DA of IPv6 packet
set to A100::. This makes the probe loops back to the centralized
monitoring system.
In our reference topology in Figure 1, N100 uses an IGP protocol
like OSPF or ISIS to get the topology view within the IGP domain.
N100 can also use BGP-LS to get the complete view of an inter-domain
topology. In other words, the controller leverages the visibility of
the topology to monitor the paths between the various endpoints
without control plane intervention required at the monitored nodes.
4. Security Considerations
This document does not define any new protocol extensions and relies
on existing procedures defined for ICMP. This document does not
impose any additional security challenges to be considered beyond
security considerations described in [RFC4884], [RFC4443], [RFC792]
and RFCs that updates these RFCs.
5. IANA Considerations
This document does not define any new protocol or any extension to
an existing protocol.
6. References
6.1. Normative References
[RFC4884] Extended ICMP to Support Multi-Part Messages. R. Bonica,
D. Gan, D. Tappan, C. Pignataro. April 2007.
[RFC4443] Internet Control Message Protocol (ICMPv6) for the
Internet Protocol Version 6 (IPv6) Specification. A.
Conta, S. Deering, M. Gupta, Ed. March 2006.
[RFC792] Internet Control Message Protocol. J. Postel. September
1981.
ali, et al. Expires June 23, 2018 [Page 20]
Internet-Draft SRv6 OAM December 2017
[RFC5837] Extending ICMP for Interface and Next-Hop Identification.
A. Atlas, Ed., R. Bonica, Ed., C. Pignataro, Ed., N. Shen,
JR. Rivers. April 2010.
[RFC7880] Seamless Bidirectional Forwarding Detection (S-BFD).
C.Pignataro, D.Ward, N.Akiya, M.Bhatia, S.Pallagatti. July
2016.
[RFC7881] Seamless Bidirectional Forwarding Detection (S-BFD) for
IPv4, IPv6, and MPLS. C.Pignataro, D.Ward, N.Akiya. July
2016.
[I.D-filsfils-spring-srv6-network-programming] SRv6 Network
Programming, draft-filsfils-spring-srv6-network-
programming, C. Fisfils, work in progress.
6.2. Informative References
[I.D-draft-ietf-spring-oam-usecase] A Scalable and Topology-Aware
MPLS Dataplane Monitoring System. R. Geib, C. Filsfils, C.
Pignataro, N. Kumar, work in progress.
[I-D.brockners-inband-oam-data] Data Formats for In-situ OAM. F.
Brockners, work in progress.
[I-D.brockners-inband-oam-transport] Encapsulations for In-situ OAM
Data, F.Brockners, work in progress.
[I-D.brockners-inband-oam-requirements] Requirements for In-situ
OAM, F.Brockners, work in progress.
7. Acknowledgments
To be added.
ali, et al. Expires June 23, 2018 [Page 21]
Internet-Draft SRv6 OAM December 2017
Authors' Addresses
Clarence Filsfils
<Cisco Systems, Inc.>
Email: cfilsfil@cisco.com
Zafar Ali
Cisco Systems, Inc.
Email: zali@cisco.com
Nagendra Kumar
Cisco Systems, Inc.
Email: naikumar@cisco.com
Carlos Pignataro
Cisco Systems, Inc.
Email: cpignata@cisco.com
Faisal Iqbal
Cisco Systems, Inc.
Email: faiqbal@cisco.com
John Leddy
Comcast
Email: John_Leddy@cable.comcast.com
Robert Raszuk
Bloomberg LP
731 Lexington Ave
New York City, NY10022, USA
Email: robert@raszuk.net
Satoru Matsushima
SoftBank
Japan
Email: satoru.matsushima@g.softbank.co.jp
Bart Peirens
Proximus
Email: bart.peirens@proximus.com
Gaurav Naik
Drexel University
United States of America
Email: gn@drexel.edu
ali, et al. Expires June 23, 2018 [Page 22]