Hi there,
A brief overview of what is happening in the IETF in working groups related to routing. This is not necessary a complete list, and given the proximity of upcoming IETF 103 meeting, there will be a large number of new submissions and changes to existing documents in the coming weeks. This list is built based on the state of WGs and documents just prior to RIPE 77 meeting. The goal is to provide an informal description of what is happening on the protocol and technology development, and what eventually may come to operations communities in not too distant future.
The
intention is to draw attention of RIPE Routing WG
participants onto what is in
development in the IETF, with a particular wish to bring in
the comments and
feedback from RIPE community into the IETF. If you see a
topic being of
interest and relevant to your environment, please consider
reviewing the
documents and providing feedback, either to the WG mailing
list, to the
document authors, or to WG chairs. If you would be
interested in becoming more involved in IETF document review
but either hesitate or are not certain what that might look
like and what the time commitment would be - please reach to
me either in the hallway or via mail.
BESS
BESS WG covers connectivity services that use BGP in one or
another form for their operation – L3VPN, L2VPN, EVPN, and
numerous their derivatives. At the current time the major
focus is on EVPN features and extensibility, the rest of the
topics are in maintenance mode.
- EVPN maintenance. Multi-attachment operation, optimizations
of multipoint frame distribution, service redundancy. Reading
list: draft-ietf-bess-evpn-bum-procedure-updates,
draft-ietf-bess-evpn-inter-subnet-forwarding.
- L2VPN and L3VPN YANG models. Configuration and monitoring
models for network elements providing both L2VPN and L3VPN
connectivity services. Not to be confused with L2 and L3
service models. Reading list: draft-ietf-bess-l3vpn-yang,
draft-ietf-bess-l2vpn-yang.
BFD
BFD WG focuses on evolution and extensions of BFD protocol and
its applicability to various dataplanes and encapsulations.
The core specification of BFD is complete and is universally
deployed, the majority of the current work is on extensions
required for new encapsulations and use cases.
- Multipoint BFD. A mechanism that allows for BFD to validate
reachability in multipoint environments. The nature of
multipoint connectivity requires the introduction of different
roles for BFD session head and session tails, it is no longer
a fully symmetric protocol. Reading list:
draft-ietf-bfd-multipoint , ietf-bfd-multipoint-active-tail.
- BFD YANG models. An NMDA compatible YANG model for managing
and monitoring BFD. Reading list: draft-ietf-bfd-yang.
- Authentication extensions. The cost of authenticating every
BFD packet makes it impractical if not completely
undeployable. However, the information that would really need
to be authenticated does not necessary repeat in every BFD
packet, therefore it may be enough to authenticate only those
BFD packets that signal a state change. Another extension is
for a change on how sequence numbers are generated. Reading
list: draft-ietf-bfd-optimizing-authentication,
draft-ietf-bfd-secure-sequence-numbers.
- Unsolicited BFD. A seemingly simpler mechanism than Seamless
BFD for establishing a BFD session to an endpoint that has no
specific initial session configuration. Reading list:
draft-chen-bfd-unsolicited.
BIER
BIER WG focuses on defining a new dataplane encapsulation and
a new forwarding paradigm for multipoint distribution of
packets – ie, redefining multicast forwarding in a scalable
and deterministic way. The core of BIER specification is
complete, current work focuses on various extensions and
integration with other technology components, as well as
looking into usability and deployment aspects of it.
- BIER specification. The core BIER specification is complete.
There are extensions for various control plane components that
need to be finalized. Reading list: RFC8279, RFC8296,
draft-ietf-bier-evpn, draft-ietf-bier-path-mtu-discovery.
- BIER integration. Not all BIER deployments will be
greenfield, there is a need for interoperability mechanisms
with existing multicast environments. Reading list:
draft-ietf-bier-use-cases,
draft-przygienda-bier-migration-options,
draft-ietf-bier-pim-signaling.
- BIER manageability. YANG model for configuration and
monitoring of BIER nodes. Reading list:
draft-ietf-bier-bier-yang.
CCAMP
CCAMP WG works on defining a set of control plane components
for non-packet based, predominantly optical transport
networks. This is a long lived WG, having completed most of
the core protocol work a while ago, with the current focus
being on maintenance and extensions, and a larger amount of
work on manageability aspects of transport networks.
- Modelling. YANG models for transport network topologies and
individual connectivity features. Reading list:
draft-ietf-ccamp-wson-yang, draft-ietf-ccamp-l1csm-yang,
draft-ietf-ccamp-alarm-module.
DETNET
DETNET WG addresses the problem of deterministic operation in
packet based networks where packet delivery timing and
reliability have strict bounds. A relatively new WG, looking
into the applicability of existing IETF technologies to time
sensitive data communication systems within a bounded domain.
- Overall architecture. Definition of main assumptions of how
deterministic traffic flows could be mapped onto packet
network infrastructure, what would the requirements be, and
how those requirements could be fulfilled. Reading list:
draft-ietf-detnet-architecture,
draft-ietf-detnet-problem-statement,
draft-ietf-detnet-use-cases.
- IP and MPLS dataplanes. How IP, and MPLS and its derivatives
could be applied to transport time sensitive payloads, what
would be the assumptions and limitations made, and what is
required for extensibility. Reading list:
draft-ietf-detnet-dp-sol-ip , draft-ietf-detnet-dp-sol-mpls.
- Bounded latency. A subset of new work on characterizing the
total end to end latency and amount of required resources for
zero loss transmission of flows. Reading list:
draft-geng-detnet-requirements-bounded-latency,
draft-finn-detnet-bounded-latency,
I2RS
I2RS WG works on defining programmatic interfaces to routing
system – a way of accessing and manipulating routing
information by ways other than via routing protocols. Most of
the chartered work is done, there is little activity, the
working group is approaching its closure, and, sadly, little
of practical applicability exists in the developed mechanisms.
IDR
IDR WG focuses on the protocol that glues the Internet
together – BGP. While the core specification of BGP has been
stable for a long time, there is a large amount of extensions
and new features being worked on.
- Extended messages. Initially BGPsec signed path attributes,
and nowadays new address families and general update packing
efficiency are being affected by a limit of a BGP message size
of 4096 octets. A simple mechanism allows for extending BGP
messages to carry payload of up to 64K octets. Reading list:
draft-ietf-idr-bgp-extended-messages.
- Flowspec maintenance. A set of clarifications and fixes to
the base Flowspec document RFC5575, and various optimizations
for traffic matching and redirect operations. Reading list:
draft-ietf-idr-rfc5575bis,
draft-ietf-idr-flowspec-interfaceset,
draft-ietf-idr-flowspec-path-redirect,
draft-khare-idr-bgp-flowspec-payload-match.
- RS BFD. A mechanism for BFD based dataplane connectivity
validation, especially in route server environments. Reading
list: draft-ietf-idr-rs-bfd.
- New BGP communities. There are still not enough of different
types of BGP communities. :-) Extended communities are not
fully AS4 capable, and various VPN address families will
require that. Reading list:
draft-heitz-idr-extra-extended-community,
draft-ietf-idr-wide-bgp-communities.
- BGP autodiscovery. There appears to be an increased interest
in various forms of BGP neighbour and topology discovery
mechanisms, mostly targeted to densely interconnected
topologies. Reading list:
draft-raszuk-idr-bgp-auto-session-setup,
draft-xu-idr-neighbor-autodiscovery,
draft-acee-idr-lldp-peer-discovery.
- Optimal reflection. Route reflectors help with overall
routing information state reduction but at a cost of topology
hiding, and providing one set of best paths for all clients.
An extension to reflector operation allows for different
clients to receive different sets of best paths. Reading list:
draft-ietf-idr-bgp-optimal-route-reflection.
- Route refresh extensions: Route refresh is a widely deployed
mechanism and is considered to be a best practice. An
extension allows for a more granular request for refreshing a
specific address family or a set of prefixes or route types.
Reading list: draft-idr-bgp-route-refresh-options.
- BGP compression. The increase of the amount of state that
needs to be transferred between the peers directly affects the
amount of octets that need to be sent and received. Observing
that there is a large number of repetitive patterns in BGP
payload, a mechanism to bundle together and compress BGP
messages is defined. Reading list:
draft-przygienda-idr-compressed-updates.
- Modelling. A YANG model for configuration of both BGP
sessions and BGP routing policy. Reading list:
draft-ietf-idr-bgp-model.
LISP
LISP WG works on a routing and addressing architecture that
separates routing and node identification namespaces and allow
for routing system scalability to extend substantially beyond
current single namespace approach. Majority of core protocol
work is complete, current work focuses on extensibility and
implementation of LISP based connectivity services.
- Core protocol maintenance. Maintenance and extensions of
core LISP architectural model and control plane components.
Reading list: draft-ietf-lisp-rfc6830bis,
draft-ietf-lisp-rfc6833bis
- Modelling. LISP YANG model for configuration and management.
Reading list: draft-ietf-lisp-yang.
- Mapping system evolution. Reading list:
draft-ietf-lisp-pubsub.
- Mapping system security. A set of mechanisms for ensuring
the authentication and integrity of mapping system
information. Reading list: draft-ietf-lisp-sec.
LSR
LSR WG focuses on evolution of IS-IS and OSPF routing
protocols, it is a merger of previously separate ISIS and OSPF
WGs. Majority of the work is on extensions needed for other
WGs, with a new work gaining traction in defining a more
scalable IGP operation in densely interconnected environments.
- IS-IS over TCP. IGPs generally use their own transport
mechanisms and as a result the complexity required for
ensuring robust and deterministic behaviour of an IGP during
the periods of synchronization of large amounts of state is
both nontrivial and proprietary. Offloading a large part of
that complexity to underlying reliable transport may seem to
be a practical way forward, and a proposed mechanism uses TCP
for point-to-point dissemination of state between IS-IS
routers. Reading list: draft-hsmit-lsr-isis-flooding-over-tcp.
- IGP flooding optimizations. A set of proposed mechanisms to
limit the over-redundant flooding of the topology information
distribution by constraining a set of links over which it gets
propagated. Reading list: draft-li-dynamic-flooding,
draft-cc-isis-flooding-reduction.
- Flexible IGP algorithms. A mechanism to define a
user-specified topology calculation methodology with a set of
metric types and constraints for building other than shortest
metric based topologies. This allows for an IGP to calculate
the specific topology itself and not rely on the external TE
component. Reading list: draft-ietf-lsr-flex-algo.
- Topology scalability. A quarter of century ago ATM PNNI
defined multilevel routing protocol loosely based on OSPF for
addressing topology information overload problems, and that
worked. Current developments for adjusting IGPs to operate
well in large and densely interconnected meshes start to look
into a similar direction by increasing the number of topology
levels upward from two. Reading list:
draft-li-hierarchical-isis, draft-li-area-abstraction.
LSVR
LSVR WG is a new working group focusing on development if a
purpose-built routing protocol based on BGP components and
targeted for a densely interconnected networks, with a
particular focus to DC use cases.
- Problem space. Definition of a problem to be solved and
reasoning why a new routing protocol would be of value and
need in the context of DC routing. Reading list:
draft-ietf-lsvr-applicability. :
- BGP SPF. The mechanics of operation of a proposed link state
vector routing protocol. In essence, BGP transport intermixed
with SPF topology calculation. Reading list:
draft-ietf-lsvr-bgp-spf.
MANET
MANET WG focuses on developing of routing protocols for
environments exhibiting many changes in topology and link
availability.
- DLEP extensions. Reading list:
draft-ietf-manet-dlep-credit-flow-control,
draft-ietf-manet-dlep-traffic-classification,
draft-ietf-manet-dlep-multi-hop-extension,
draft-ietf-manet-dlep-lid-extension.
MPLS
MPLS WG coordinates the development of MPLS dataplane
encapsulation and control plane components required for its
operation. Being an active WG for a third decade now, majority
of core protocol work is complete, the focus is on maintenance
and extensions required for adjacent uses of MPLS derivative
technologies, and manageability.
- MPLS modelling. Definition of YANG models for configuration
and operation of MPLS functionality. Reading list:
draft-ietf-mpls-base-yang, draft-ietf-mpls-static-yang.
NVO3
NVO3 WG works on IP based encapsulations for network
virtualization, with the focus on DC environments.
- Geneve. VXLAN encapsulation, while not a product of an IETF
WG, proved to be successful in the industry, reaching a wide
universal deployment. As the requirements for encapsulation
increased in order to fulfil different use cases, it became
evident that VXLAN has fundamental limitations and a new
encapsulation protocol is required. The work resulted in
selecting Geneve as a successor to VXLAN, providing practical
extensibility, hardware implementation efficiency, and
security awareness. The work on Geneve is nearing completion.
Reading list: draft-ietf-nvo3-encap, draft-ietf-nvo3-geneve.
- EVPN control plane for NVO3. While NVO3 WG is not directly
chartered to work on control plane components, practical
deployments will be and are based on some form of control
plane. EVPN is one possible candidate of such control plane,
providing autodiscovery of participating NVEs and the
distribution of addresses and their mappings. Reading list:
draft-ietf-nvo3-evpn-applicability,
draft-boutros-bess-evpn-geneve.
- NVO3 security. An analysis of security requirements both for
the NVO3 environment and NVO3 dataplane encapsulators. Reading
list: draft-mglt-nvo3-geneve-security-requirements.
PALS
PALS WG works on maintenance of LDP signalled services that
use pseudowires.
- Ethernet PW CW. Mandating the use of control word for
Ethernet pseudowires in order to avoid the possible aliasing
problems based on inability to deterministically distinguish
payload data from encapsulated frame header. Reading list:
draft-ietf-pals-ethernet-cw.
- Maintenance and extensions. The WG mostly focuses on
extensions of existing signalling mechanisms and services,
with nothing exceptional to report at this time.
PCE
PCE WG works on defining mechanisms for centralized topology
computation for MPLS-derivative networks.
- Applicability to Segment Routing. Reading list:
draft-ietf-pce-segment-routing.
- PCE modelling and manageability. Reading list:
draft-ietf-pce-pcep-yang.
PIM
PIM WG works on development of IP multicast routing and
signalling control plane components.
- YANG models for IP multicast. Configuration and maintenance
models for IP multicast control protocols and network element
operation. Reading list: draft-ietf-pim-msdp-yang,
draft-ietf-pim-igmp-mld-yang.
- Protocol maintenance and extensions. One of the aspects
related to PIM stability on multi-access networks is the
proper election and maintenance of a DR. Trying to rely on the
implicit expiration of DR timers and switching over only then
does not necessarily lead to fast convergence, and does not
allow for a deterministic selection of backup DR in advance. A
proposed extension allows for carrying information about
candidate DR nodes and makes a way for off-PIM liveness
validation. Reading list: draft-ietf-pim-dr-improvement.
RIFT
RIFT WG works on a specialised routing protocol having a
mixture of link-state and distance-vector operations, targeted
for a specific Clos-based topologies that exhibit a large
number of interconnections between points in the network.
- Base specification. As for a relatively new WG, the current
main focus is on base specification, including protocol
mechanics and encodings, and targeted use cases. Reading list:
draft-ietf-rift-rift, draft-filyurin-rift-access-networks.
- Modelling. A configuration and operation YANG model for RIFT
nodes. Reading list: draft-zhang-rift-yang.
ROLL
ROLL WG works on development of routing protocols for low
power and lossy networks and nodes that have limited
communication resources.
- RPL applicability. A view on applicability of RPL routing
protocol to low power and lossy networks. Reading list:
draft-ietf-roll-rpl-observations.
RTGWG
RTGWG WG focuses on new and innovative topics related to
routing that are either too wide for a specific existing
subject WG at the time, or too narrow and not fitting any
particular WG. It acts as an initial filter and redirection
mechanism for new work coming into routing area.
- BGP PIC. A set of mechanisms and recommendations on how a
node can implement routing and forwarding logical data
structures in order to limit the scope of changes needed to be
done on the node related to the changes in routing topology,
particularly targeting BGP environment. Reading list:
draft-ietf-rtgwg-bgp-pic.
- LNE and NI models. A set of YANG models describing the
partitioning of a network element into instances or VRFs and
how it can be represented as a hierarchy of manageable
entities from the control plane component perspective. Reading
list:
- Routing policy modelling. A YANG model defining how a local
network element routing policy (not the BGP routing policy)
could be represented and used. Reading list:
SFC
SFC WG works on definition of an overall architecture and
technology components for service chaining.
- SFC OAM. A set of mechanisms for allowing to monitor and
validate the operation of SFC environments. Reading list:
draft-ietf-sfc-oam-framework.
- SFC and In-situ OAM integration. Reading list:
draft-ietf-sfc-ioam-nsh, draft-ietf-sfc-proof-of-transit.
SPRING
SPRING WG focuses on overall architectural development of
Segment Routing and its associated protocols and mechanisms,
as well as focusing on operational and usability aspects of
it.
- SR MPLS dataplane. The components and protocol mechanics for
Segment Routing using MPLS dataplane. Reading list:
draft-ietf-spring-segment-routing-mpls.
- SR policy routing. A framework on how SR components could be
bound together and used for implementation of a scalable
source based routing mechanism. Reading list:
draft-ietf-spring-segment-routing-policy.
TEAS
TEAS WG focuses on aspects related to traffic engineering
technologies for MPLS and derivative dataplanes and signalling
components. The core set of specifications has been completed,
current work focuses on extensions and feature maintenance,
and on development of YANG models.
- TE topology modelling. Reading list:
draft-ietf-teas-yang-te, draft-ietf-teas-yang-te-topo,
draft-ietf-teas-te-topo-and-tunnel-modeling.
- Enhanced VPN. A feasibility study on what IETF technology
components could be reused and what needs to be developed for
VPNs that would allow for stricter resource guarantees and
awareness to network applications. Reading list:
draft-dong-teas-enhanced-vpn.
MBONED
MBONED WG focuses on deployment and operation aspects of
multicast technologies and networks.
- YANG model for multicast configuration and monitoring.
Reading list: draft-ietf-mboned-multicast-yang-model.
- Multicast in DC environments. Multicast traffic especially
in DC environments has been frowned upon for a long time, and
the typical application communication pattern was
predominantly point to point. Both of those assumptions start
to change, and an analysis of what the effect of bringing in
multicast at scale into DC environment would be needs to be
performed. Reading list: draft-ietf-mboned-dc-deploy.
- Multicast in wireless environments. IEEE 802 wireless is not
the mostly friendly environment for IP multicast.
Recommendations both for deployment, and for cross-SDO
development of wireless and multicast technologies, are
analysed. Reading list:
draft-ietf-mboned-ieee802-mcast-problems.
GROW
GROW WG focuses on operational aspects of global BGP routing
system and works on toolkits, methodologies, and protocols
targeted for BGP operation and maintenance.
- BMP maintenance. BMP is starting to get operational traction
and thus limitations and shortcomings of the original design
show up. The current work focuses on allowing BMP to be able
to export information about local and outgoing RIBs, and a
rather large item of work on extending the format and
capabilities of Route Monitoring message. Reading list:
draft-ietf-grow-bmp-adj-rib-out,
draft-ietf-grow-bmp-local-rib,
draft-hsmit-bmp-extensible-routemon-msgs.
- Actions on Well-known communities. Different BGP
implementations behave differently when setting or removing
BGP standard communities, and this may result in inconsistent
routing policy operation. The document provides the current
list of vendor behaviour, and guidelines on what vendors
should do for their implementations. Reading list:
draft-ietf-grow-wkc-behavior.
- RPKI AS Cones. A mechanism for aggregation of a list of
client AS numbers for purposes of RPKI validation. Reading
list: draft-ietf-grow-rpki-as-cones.
SIDR
SIDR WG has completed its chartered work and has been closed
recently. The main deliverables of SIDR are BGPsec and RPKI.
- BGPsec protocol specification. Reading list:RFC8205,
RFC8206, RFC8207.
- RTR maintenance. Reading list: RFC8210, RFC8211.
SIDROPS
SIDROPS WG focuses on deployment and operational aspects of
global routing system security mechanisms by providing
guidance on how protocols and mechanisms developed in SIDG WG
couls and should be deployed and used, and acting as a
distributor of operational knowledge.
- TAL HTTPS transport. An extension allowing for use of HTTPS
in addition to rsync for fetching trust anchors. Reading list:
draft-ietf-sidrops-https-tal.
- ASPA. AS Provider Authorization, a simple yet practical way
of checking the validity of an AS path. Reading list:
draft-azimov-sidrops-aspa-profile,
draft-azimov-sidrops-aspa-verification.
OPSAWG
A WG for generic operations related topics that do not have
any other more suitable or more specific WG.
- TACACS. The documentation of the current deployed TACACS
protocol, with the explicit goal of not defining new features
and extensions but documenting the current state of the
implemented and operated protocol. The specification carries
more of a historical and informational value of what has been
implemented. Reading list: draft-ietf-opsawg-tacacs. There is
another document on defining YANG model for TACACS, it is an
early start of the work and the scope is likely too narrow and
too focused on TACACS only instead of having a broader scope
of extending RFC7317 system authentication model to cover
multiple authentication, authorization, and accounting
mechanisms. Reading list: draft-zheng-opsawg-tacacs-yang.
- IPFIX flow export based on BGP communities. Similar to
NetFlow/IPFIX export of AS numbers and IP addresses, there may
be use cases where a more granular and user-controlled marking
of flows might be needed. The proposal defines a set of IPFIX
information elements and a mechanism for exporting flows based
on BGP communities of various types. Reading list:
draft-ietf-opsawg-ipfix-bgp-community.
- NAT modelling. A YANG model for configuration of various
forms of NATs, providing a flexible framework for possible
future additions of new NATs. Reading list:
draft-ietf-opsawg-nat-yang.
- Secure Device Install. A discussion in early stages on a
simple proposal of zero-touch initial provisioning mechanism
without requirement for remote hands intervention into network
element configuration. With the cooperation of network element
manufacturer, the initial configuration is encrypted with the
public key of the node, thus locking the configuration to the
intended target node and protecting it from the third parties
that may be involved in the initial provisioning process.
Reading list: draft-wkumari-opsawg-sdi.
KARP
KARP WG was focusing on transport security aspects of routing
protocols, and has been closed a while ago due to the lack of
energy. There seems to be a growing interest in the community
on revisiting some of the work items of KARP, especially on
developing a successor mechanism to TCP MD5 authentication
signatures.
- Routing transport security. TCP-AO has been around for a
long time, however, there are no deployments as there are no
practical implementations. Reading list: RFC5925.
- Using IKE for routing protocols. This is a “homeless”
document at this time, but worth reading and commenting on.
The overall approach is quite simple – IKE is meant for
distributing keying material, and if instead of IPsec
components the users of that keying material would be routing
protocols – could this work? It appears that it could, the
overall solution is not that complex. Reading list:
draft-mahesh-karp-rkmp.
NETCONF
NETCONF WG focuses on development of protocols components for
evolving model based network management approaches. The core
of both NETCONF and RESTCONF protocol work is complete, the
focus is on extensions and adjustment to operational
requirements as a feedback to deployments.
- Event notifications and model based telemetry. Mechanisms
for receiving events and data from network elements, as
opposed to sending configuration and state into network
elements. Reading list:
draft-ietf-netconf-netconf-event-notifications,
draft-ietf-netconf-yang-push.
- Zero-touch provisioning. A set of mechanisms for initial
bringup of network element.Reading list:
draft-ietf-netconf-zerotouch.
NETMOD
NETMOD WG works on evolution of YANG modelling language and
overall modelling methodology, as well as working on “core”
set of models.
- Datastore architecture. An architectural framework for
allowing the configuration to be represented and interpreted
in more than a single context, reflecting the operational
reality of having the target configuration not necessary
corresponding to the actual operating parameters. Reading
list: RFC8342, draft-ietf-netmod-schema-mount.
L2SM, L3SM
L2SM and L3SM WGs were a short lived and narrow focused on
developing the service model for L2VPN and L3VPN connectivity
services – contrary to the work happening in BESS WG on
element level configuration models for L2VPN and L3VPN.
Reading list: RFC8049, RFC8466.
IPPM
IPPM WG works on measurement instrumentation – defining
metrics, methodologies, and protocol extensions for
measurement of IP network performance.
- In-situ OAM. A set of mechanisms and encapsulation
extensions that allow for recording of the measurements that
are applicable to the transit packet within the packet itself.
Reading list: draft-ietf-ippm-ioam-data ,
draft-brockners-ippm-ioam-geneve.
- AURA. A multipath-aware measurement methodology and
mechanisms. Reading list: draft-ietf-ippm-route.
Other
interesting assorted reading materials
- FAST. An approach to coordination of and awareness to
bidirectional routing. Reading list: draft-herbert-route-fast.
- Protocol wire image. A methodology for analysing of
visibility of protocol signatures and inferred working to a
side observer. Reading list: draft-iab-wire-image.
- Path signals. A set of analysis and recommendations on side
effects of increasing use of encryption in packet networks.
Reading list: draft-iab-path-signals.