YANG Takes Off in the Industry

In 2003, RFC 3535, “Overview of the 2002 IAB Network Management Workshop” documented the outcomes of a dialog started between network operators and protocol developers to guide the IETFs focus on future work regarding network management. The 14 operator documented requirements led to the sequential creation of the NETCONF Working Group the same year, the NETMOD Working Group in 2008, and the core data models. The work resulted in XML-based Network Configuration Protocol (NETCONF) RFCs 6241, 6242, 6243, and 6244, in 2011 (respectively revised from 4741, 4742, 4743, 4744) and the associated data modeling language YANG RFCs 6020 and 6021 in 2010. Unfortunately, the deployment of new management protocols takes longer compared to non-management protocols. Indeed, it takes a while to get those RFCs implemented in the majority of devices. But now we see that the new protocol and data modeling language are picking up.

At the IETF 90 last week, starting on Sunday, the “YANG Advice and Editing Session” took place. A “pyang” tutorial, presented by Ladislav Lhotka, and a presentation on code generation experience from the OpenDaylight project started the afternoon (both sessions were recorded and will be shared). Then, during 2h30, the YANG doctors provided their advice on existing YANG models, both individual and WG documents. In total, 13 YANG models were reviewed, and some updated on the fly during the afternoon. This experience, which received very good feedback, should speed up the development of standardized YANG models in the industry. The YANG doctors found this session useful as well, for their update of “Guidelines for Authors and Reviewers of YANG Data Model Documents” RFC 6087bis. We plan on repeating this “YANGathon” sesion in future IETF meetings.

At this IETF, “YANG” was the subject of many discussions. The I2RS WG selected YANG for their work a couple of weeks ago. The routing experts discussed OSPF, ISIS, BGP models, all of them based on the foundational “YANG Data Model for Routing Management”. YANG was mentioned as potential solutions in BoFs, for example the Abstraction and Control of Transport Networks (ACTN), as a way to specify service models. And also in side meetings such as “Application-based Policy for Network Functions” (APONF). Confirmed by hallway conversations, we can expect way more YANG models in the future. This is a good problem, but also one we have to prepare for.

Is this YANG interest only in the IETF? Not quite. We see a lot of interest for YANG models in the Open DayLight open source project, with more than 100 YANG models. At the last IEEE meeting, Andy Bierman provided a tutorial on NETCONF and YANG, targeted for people who have been using SNMP and MIB modules. This will certainly trigger a process of transformation in the IEEE 802 management space. Regarding MIB modules, let me stress one more time the “Writable MIB Module IESG Statement” IESG statement, which explains that SNMP is not ideal for configuration of devices, while NETCONF is.

On the tools side, the NETMOD WG is experimenting with using github for faster data model development. So far, a github repository has been created that a number of YANG models from standards organizations such as the IETF, open source such as the Open DayLight project, or vendor proprietary ones. After the “YANG Advice and Editing Session” on Sunday, some models were directly updated in github. This repository should not only speed up data model iterations and developments but also brings bridge the gap between two communities, the IETF community and the open source community.

To the question “Where should the YANG models be specified in the IETF?”, discussed on regularly, the answer is: The end goal is that the different Working Groups produce their own YANG models. For example, the “6TiSCH: “IPv6 over the TSCH mode of IEEE 802.15.4e” charter mentions data models. The YANG doctors reviewed two 6tisch WG documents during the “YANG Advice and Editing Session“, guiding the authors in the right direction.

However, not all WG charters consider data models at this point in time. Note the recent change in the NETMOD charter:

The NETMOD WG may also develop any additional data models written in YANG that the WG considers core building blocks and that do not fall under the charters of other active IETF working groups. The NETMOD WG may simply add such milestones as it sees fit, with the advice and consent of the responsible AD.

Finally, the NETMOD WG scheduled weekly calls to go over the YANG 1.1 issues. With all the energy around NETCONF and YANG, and the willingness to work more efficiently and faster, a physical NETMOD interim is planned for September (see the NETMOD mailing list for the details).

All these observations point to a single conclusion: YANG takes off in the industry! The YANG community is dedicated to hep make YANG a success.

Benoit Claise, OPS Area Director