Markdown Version | Session Recording

Session Date/Time: 16 Jun 2025 14:00

GREEN

Summary

The GREEN Working Group held an interim meeting to discuss updates on existing drafts, primarily focusing on the framework for energy-efficient management and proposed Yang data models. A significant portion of the meeting was dedicated to an in-depth discussion of challenges and open questions related to data model design, including power control granularity, aggregation, and the relationship between models and the framework. The chairs proposed forming a data model design team to accelerate progress, and this proposal received strong support from those present. An upcoming hackathon focused on green networking use cases was also announced.

Key Discussion Points

• Framework for Energy-Efficient Management (draft-bell-green-framework):

  • The framework proposes three levels (Network Domain, Controller, Device/Component) and three key interfaces (Inventory Capability, Monitoring, Control).
  • Core concepts include Energy Objects, Metering Relationships (X meters for Y), and Power Source Relationships (X powers Y), with a brief mention of Aggregation Relationships. These relationships are crucial for understanding energy flows and avoiding double-counting in measurements.
  • An open question was raised regarding the granularity of Power State Sets (e.g., simple on/off vs. more complex states like hibernate, sleep, or minimum efficiency). Standardizing these states across diverse devices and vendors presents a challenge, particularly when linking them to performance SLAs (e.g., recovery time, throughput, latency).
  • Discussion highlighted the need for data models to support both hiding complexity for mainline cases while retaining auditable detail for scenarios like customer/use case assignment.

• Yang Data Models for Energy Efficiency:

  • The presentation outlined a proposed model architecture with Network Inventory, Network Level Energy Efficiency Management, and Device Level models.
  • Existing work (e.g., draft-cwb-green-energy-saving-management and draft-li-green-power) was referenced, with a proposal to consolidate common definitions into an energy-efficiency-common module.
  • This common module would define features, identities (energy saving modes, methods, power states), types, and groupings for energy consumption data and power parameters.
  • Open issues in data model design:
    • Framework Dependence: Whether data models should be tightly coupled with the framework or remain largely independent, with a consensus that core Yang keys (like UUIDs for devices/components) should be aligned.
    • Model Layering: The need for device/component level, network level (inventory/topology), and potentially higher-level service models to allocate energy consumption to specific services or customers.
    • Power Control Granularity: Whether control states should be generic (e.g., EMAN's work) or specific to component types (e.g., line cards, optics), and how to align configurable states with operational states.
    • Control Intent: Whether control actions relate to desired operational intent (e.g., "sleep capable mode") allowing the device to optimize, or direct, granular power usage control.
    • Aggregation: How to model aggregation at device and network levels, report aggregate and fine-grained data without double-counting.
    • Reporting Accuracy/Granularity: Whether measurement accuracy should be statically defined in the model or dynamically reported by devices.
    • Time Intervals: How to standardize or manage time intervals for aggregated power usage reporting.

• Use Cases and Terminology Documents:

  • No updates were presented for the use cases document, with a goal for adoption by the next meeting.
  • A request for Working Group adoption of a terminology document was noted, with discussion deferred to IETF 123.

• Hackathon Announcement (IETF 123 Madrid):

  • An upcoming hackathon will exercise green networking use cases, including applying the green framework, collecting energy information from cloud-native functions, end-to-end metrics, video streaming energy consumption, and addressing double-counting with smart PDUs.
  • The setup will involve an Exigence streaming environment, wireless access points, physical meters, and access to smart PDUs in Telefonica labs. Data collection and visualization will be key.

Decisions and Action Items

• The request for a Working Group adoption call on the common-energy-efficiency-model module was deferred. The chairs indicated that more alignment on design direction is needed first.
• Discussion on the Working Group adoption of the terminology document was deferred to IETF 123. Participants were encouraged to review the document in the meantime.
• A poll of the room indicated strong support for forming a Data Model Design Team (12 votes for, 2 against). The chairs will discuss how to proceed based on this feedback.
• A poll of the room indicated strong support for arranging more frequent interim meetings (12 votes for, 1 against).
• The chairs will discuss the feedback received on the design team and more frequent interim meetings to determine the best approach for progressing the data model work.
• Luis will send an email to the GREEN WG mailing list advertising the upcoming hackathon and inviting participation.

Next Steps

• The chairs will evaluate the feedback on a Data Model Design Team and increased interim meeting frequency, and communicate a plan for accelerating progress on the Yang data models.
• Working Group members are encouraged to review the terminology documents in preparation for discussion at IETF 123.
• Participants interested in contributing to the hackathon at IETF 123 should reach out to Luis and monitor the mailing list for updates.
• A call for presentations for the GREEN sessions at IETF 123 will be announced.