GREEN

Summary

The GREEN working group met at IETF 126 to discuss the progress of its core deliverables, evaluate newly adopted works, and consider related drafts for potential adoption or cross-working-group collaboration. The session was chaired by Rob Wilton and Diego Lopez. Excellent progress has been made since IETF 125, with several core documents now adopted. The group discussed alignments across terminology, use cases, and the framework, finalized key updates on the base device YANG model, and initiated structural scoping for a controller-level YANG model. Additionally, experimental results from the recent Hackathon were shared, and several new proposals in areas such as telemetry provenance, IPFIX energy consumption, and multicast eco-modes were discussed.

Key Discussion Points

1. Introduction & WG Status

Presenters: Rob Wilton & Diego Lopez
Slides: IETF 126 GREEN Chairs Slides
The chairs noted that the core documents—covering use cases, terminology, the framework, and the base device YANG model—have all been adopted.
Interim Meetings: Bi-weekly design meetings (lightweight design team format) are currently held every other week around 2:00 PM UTC/BST. The chairs encouraged more active participation and invited feedback if the timing is inconvenient for members in other time zones.
Coordination: The WG is actively coordinating with LSR (e.g., Tony Li) to ensure that energy-related YANG models in routing groups do not define incompatible terms or duplicate work.
Liaisons: Discussion during the adoption call for the framework highlighted European Commission and other SDO activities regarding real-time energy measurement legislation. Diego Lopez volunteered to draft an outgoing liaison statement to raise awareness of GREEN's standardization efforts.

2. Terminology for Energy Efficiency Network Management

Presenter: Qin Wu
Slides: Terminology for Energy Efficiency Network Management
Draft: draft-ietf-green-terminology (v02)
Qin Wu outlined updates aiming to align the terminology draft with the framework and YANG modules.
Key changes in v02 include:
- Moving "power factor" and "energy management system" definitions from the framework draft to the terminology draft.
- Adding definitions for "nameplate power" (referencing RFC 7326) and "power source".
- Refining the "power factor" definition (defined as a percentage value between 0 and 1, defaulting to 1 for direct current power sources).
- Modifying the "energy management system" definition to remove a note regarding carbon reporting that fell outside the current charter.
Action: The individual repository will be formally moved to the GREEN WG GitHub organization.

3. IETF126 use cases draft 01

Presenter: Emile Stephan
Slides: IETF126 use cases draft 01
Draft: draft-ietf-green-use-cases
Emile Stephan reported that the use-cases document is being kept in tight alignment with the modeling team. The immediate focus is on ensuring support for short-term use cases such as incremental deployment and double accounting.
The authors aim to evaluate real-world implementations of these use cases prior to IETF 127.
There was a brief discussion on whether this draft needs to be formally published as an RFC or kept as a living reference. Rob Wilton suggested a future working group last call to solidify consensus on the covered use cases, while Diego Lopez reminded the authors to refresh the draft to prevent expiration.

4. GREEN framework

Presenter: Marisol Palmero
Slides: GREEN framework
Draft: draft-ietf-green-framework (v02)
Marisol Palmero presented structural cleanups in the reference model. Notably, the "main network manager" layer was renamed to "Energy Management System" (EMS) to align with older RFCs.
Key updates in v02:
- Added an EMAN migration mapping table.
- Introduced an "operational considerations" section to define inheritance rules for unit multipliers and data source accuracy to minimize telemetry payload bloat.
- Added a functional relationship type mapping: "enabled-by" and "enabling" (originally raised by Tony Li) to represent logical or hardware-hosting dependencies (e.g., a chassis hosting a line card) distinct from direct power or metering relationships.
The team will continue to focus on the alignment with device-level and future controller-level YANG models.

5. Update on YANG base model & Controller Scoping

Presenters: Benoit Claise & Jan Lindblad
Slides: Update on YANG base model -- Benoit Claise -- Jan Lindblad
Draft: draft-ietf-green-power-and-energy-yang (v03)
Benoit Claise detailed changes in the base device model, which is split into operational (read-only) and energy control (read-write) sections.
Relationship Mapping: Relationships use a string-type Peer ID instead of a UUID because routers often cannot resolve the UUID of adjacent devices natively (e.g., via LLDP or CDP).
Open Technical Issues:
- Instantaneous Power Mandatory vs. Optional: The current draft defines instantaneous power as mandatory. However, hardware constraints make it impossible or too costly to measure instantaneous power on all components. Rob Wilton and Nigel Davis suggested making the YANG leaf optional but adding conditional guidelines (e.g., "if a physical sensor is present, the node must report it").
- Data Types: Tony Li pointed out that instantaneous power is defined as an int32, which allows negative power. He suggested changing it to uint32 unless there are science-fiction use cases for negative power. Benoit Claise agreed to refine this.
- Dual vs. Single Multipliers: The team will take to the mailing list whether consumed and delivered energy require separate unit multipliers.
Controller-level YANG Model: Jan Lindblad presented on the emerging work for a controller-level YANG model (Work Item 1 & 4). The model is designed to be a Minimum Viable Draft (MVD) focusing on configuring data sources, scheduling telemetry collection, correlation of IDs between controllers and devices, data transformations (sums, averages, peaks), and exporting to external databases.
Poll:
- Should the base network controller model be scoped to include more than just power/energy attributes?
  - Yes: 16
  - No: 2
  - No opinion: 5
  - Total: 54
- Result: Strong support to scope the base network controller model beyond simple power/energy attributes. Rob Wilton advised the authors to still focus on power and energy first in the initial drafts while leaving extensible hooks.

6. PETRA Green API

Presenter: Luis Miguel Contreras Murillo
Slides: Update on draft-petra-green-api
Draft: draft-petra-green-api
Luis Miguel Contreras Murillo provided an update on the Path Energy Traffic Radio API (PETRA). PETRA acts as an Northbound Interface (NBI) from a network controller, allowing external consumers (such as SD-WAN customers) to request energy consumption statistics between specific endpoints.
Updates in this version:
- Removed out-of-scope parameters (e.g., carbon intensity) to align with GREEN's charter (such work will move to the Sustain-RG).
- Generalized the API to query by IP prefixes, MAC addresses (L2), and network slice service delivery points.
- Added traffic throughput and volume characterizing metrics.
Polls:
- Have you read/reviewed any version of the PETRA draft (draft-petra-green-api)?
  - Yes: 12
  - No: 12
  - No opinion: 0
  - Total: 55
- Do you think that the WG should adopt the petra draft at this time?
  - Yes: 11
  - No: 0
  - No opinion: 16
  - Total: 55
- Result: There was interest in adopting the PETRA draft. The chairs will take the adoption call to the mailing list, reminding the group that core deliverables remain the priority.

7. YANG Data Model for Reporting Utilization Scores in ISAC

Presenter: Mehad
Slides: A YANG Data Model for Reporting Utilization Scores in ISAC
Draft: draft-jadoon-green-isac-utilization
Mehad presented a model targeting Integrated Sensing and Communication (ISAC). ISAC components consume significant compute, power, memory, and storage resources. The proposed model reports an overall utilization score and augments the GREEN base power/energy module.
Discussion: Jan Lindblad noted that augmenting the device-level model makes sense, but cautioned that representing service-level concepts on the controller side is highly distinct. Another participant questioned how easy it is to isolate the power consumption of "sensing" versus the underlying radio access network (RAN) components, especially during idle/night periods where sensing prevents the RAN from going into deep sleep.
Poll:
- Have you read/reviewed any version of the ISAC Utilization Draft (draft-jadoon-green-isac-utilization)?
  - Yes: 6
  - No: 23
  - No opinion: 1
  - Total: 54
- Result: Since only a few people have read the draft, the chairs deferred any adoption poll and urged participants to review the draft on the mailing list.

8. GREEN Provenance

Presenter: Marisol Palmero
Slides: GREEN provenance
Draft: draft-palmero-green-provenance
Marisol Palmero introduced a draft exploring the application of telemetry provenance (signing data at the source/controller using COSE signatures) to energy and power metrics. This prevents "greenwashing" by guaranteeing that reported data is authentic, auditable, and traceably verified.
Discussion: Ben Curtis asked about verification failures (e.g., if a signature fails or is missing, should the controller drop the telemetry completely?). Mahesh Jethanandani (AD) raised a scoping question, asking if this work belongs in GREEN or if it is a general mechanism that should be standardized in the Ops Area Working Group (OPSAWG). Marisol agreed that while the signing technology is generic, the energy audit trail use case is a key driver within GREEN.

9. Multicast Eco-Mode Signaling

Presenter: Luis Miguel Contreras Murillo
Slides: draft-contreras-pim-eco-mode
Draft: draft-contreras-pim-eco-mode
Luis Miguel Contreras Murillo presented a PIM working group draft on extending IGMP/MLD signaling (RFC 9279) to allow subscribers to signal an "eco-mode" preference for multicast video streams.
The draft introduces four Eco Levels (0: No eco, 1: Low impact optimizations, 2: Moderate quality trade-offs, 3: High eco with reduced resolution/quality) and preferences like carbon awareness and renewable energy.
Discussion: Rob Wilton found the concept interesting and suggested refining the differences between Level 0 and Level 1 (since most devices optimize energy by default if there is no service impact) and simplifying the preference fields.

10. Hackathon Results

Presenter: Luis Miguel Contreras Murillo
Slides: Report on recent Hackathon results
Luis Miguel Contreras Murillo summarized experiments conducted in the Telefonica lab. Two physical routers were connected with various traffic flows (100G, 5x10G, 4x40G) to compare CLI-reported internal power measurements against external smart PDU readings.
Key Results:
- A ~6% to 10% discrepancy was observed between CLI and PDU values. David Slik suggested in the chat that this is likely because the smart PDU measures AC input power, while the router internal sensors measure DC output power after power supply unit losses.
- Idle power consumption remains very high across typical routing hardware.
- Incremental traffic added a modest but noticeable energy footprint (~10W per 100G per hop).
Future hackathons will test actual YANG Push streaming and sleep-mode transitions.

11. IPFIX Energy Consumption Export

Presenter: Jinjie Yan
Slides: draft-yan-opsawg-ipfix-energy-consumption
Draft: draft-yan-opsawg-ipfix-energy-consumption
Jinjie Yan proposed exporting energy metrics via IPFIX. The motivation is to correlate transient power spikes directly with flow metering events at the source without post-processing or polling delays. The draft's information elements are aligned semantically with draft-ietf-green-power-and-energy-yang.
Discussion: Rob Wilton expressed skepticism, noting that IPFIX also operates on configured export intervals or packet counters. He questioned why IPFIX represents a superior solution to YANG Push for capturing transient spikes. Jinjie Yan agreed to elaborate on the technical benefits and discuss this further on the mailing list.

Decisions and Action Items

Terminology Document: Move the repository of draft-ietf-green-terminology into the official GREEN WG GitHub organization (Qin Wu / Chairs).
Liaison Statement: Prepare and share draft text on the mailing list for an outgoing liaison statement informing relevant European Commission and SDO bodies of GREEN's standardization efforts (Diego Lopez).
Use Cases Refresh: Update draft-ietf-green-use-cases to prevent its expiration in the data tracker (Emile Stephan).

Next Steps

YANG Base Model Mailing List Threads:
- Initiate discussions on resolving the "instantaneous power as optional" YANG definition and incorporating descriptive profiling guidelines in the framework (Benoit Claise).
- Begin a thread to address accuracy metadata and identity refs (Jan Lindblad).
- Solicit use cases on whether consumed and delivered energy need separate unit multipliers (Benoit Claise).
Controller YANG: Progress the first MVD draft for the controller-level model (Jan Lindblad).
PETRA Adoption: Issue a formal adoption call on the mailing list for draft-petra-green-api (Chairs).
ISAC Review: Encourage the working group to review and comment on draft-jadoon-green-isac-utilization on the mailing list (All).

Automatic IETF Minutes