| Internet-Draft | legacy modularity and eco-design | July 2024 |
| Stephan, et al. | Expires 9 January 2025 | [Page] |
This document discusses the usage of inventory-maintained information for assessing the adaptation of existing devices to eco-design. This assessment is driven by the weight of the manufacturing in the sustainability cost with regard to the power consumption in operation.¶
This note is to be removed before publishing as an RFC.¶
The latest revision of this draft can be found at https://emile22.github.io/sustainability/draft-stephan-legacy-path-eco-design-latest.html. Status information for this document may be found at https://datatracker.ietf.org/doc/draft-stephan-legacy-path-eco-design/.¶
Discussion of this document takes place on the Green BoF individual mailing list (mailto:[email protected]), which is archived at https://mailarchive.ietf.org/arch/browse/green-bof/. Subscribe at https://www.ietf.org/mailman/listinfo/green-bof/.¶
Source for this draft and an issue tracker can be found at https://github.com/emile22/sustainability.¶
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). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
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."¶
This Internet-Draft will expire on 9 January 2025.¶
Copyright (c) 2024 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 (https://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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
Many companies have integrated sustainability improvements into their core business strategies. Such trend is driven by the growing awareness of environmental matters and regulatory requirements like Corporate Sustainability Reporting Directive (CSRD), a regulatory framework proposed by the European Commission to enhance corporate transparency and ensure that companies provide comparable information to assess their sustainability performance. In a close future they will integrate the regulation on ecodesign framework [(EU) 2024/1781] published in June by the European Commission.¶
Sustainability impacts numerous aspects of the life cycle management (LCM) of devices. This document discusses the benefits of leveraging existing devices modularity to introduce eco-designed components in the networks while being able to assess the gains in sustainability.¶
The motivations are the urgent need to reduce power consumption by replacing device components with more energy-efficient alternatives, and to decrease embedded carbon emissions while extending the life duration of the device chassis. It can be viewed as a very basic use case of GREEN-bof [GREEN-BOF] approach and the requirement of a fast track specification for reporting the path of a legacy device to sustainability.¶
Many Networks and networking devices (e.g., routers, switches, or Network Functions (NFs)) are designed with some modularity to cope with manufacturing and operational constraints. However, the modularity level is a still implementation and deployment specific. It is expected that legacy devices will be progressively replaced with eco-design products. Both will co-exist in networks as there is a balance to find between the increase of the live duration of existing devices and their replacement with eco-design devices [SUST-INS].¶
The same approach applies at the same time to individual devices: legacy devices will include progressively more and more eco-designed hardware components.¶
Eco-design means that the products include environmental considerations throughout their entire lifecycle to reduce their environmental impact. In France they are based on General Reference Base for Eco-Design in digital services ("RGESN") at https://ecoresponsable.numerique.gouv.fr/publications/referentiel-general-ecoconception/.¶
Eco-design can be summarized as the concepts and current practices related to the integration of environmental aspects into every stage of a product lifecycle. This starts at product design and development [ISO/TR 14062:2002 ]¶
The upgrade of legacy devices with eco-designed cards can be separate in 2 categories:¶
"Simple" update : The adaptation is simple, a card is 'just' replaced with another one which consumes significantly less power by itself when running and which required fewer resources during its manufacturing and its deployment.¶
"Complex" update: The replacement of the card requires the adding or the update of software components to enable dynamic power consumption.¶
Network operators have been updating their devices' components for decades. Consequently, replacement with eco-designed components can be begun immediately as there is no dependency on management solutions. The assessment of environmental and power gains can be done manually or with ad hoc scripts from the datasheets of the manufacturer or using an ad hoc process.¶
It is clear that information for doing static assessment is spread over many mediums or OPS interfaces (datasheet, Web URL, CLI , YANG, MIB, IPFIX ...). A proposal consists in documenting how to do static assessment for a set of devices and components based on volunteering [GREEN-BOF].¶
In the mid-term, in a way to scale the assessment, inventory [IVY-WG] requires distinguishing legacy devices which include eco-designed components and eco-designed components inside legacy devices. This must not delay the initial deployment of eco-designed components in legacy devices described above.¶
Software modularity increases with the generalisation of continuous development and deployment approaches. Power consumption of current software components of network devices is rarely evaluated. They can be updated immediately 'just' by replacing with another one which consumes significantly less power by itself.¶
It might seem inappropriate to try to decrease the power consumption of a software component as intuitively it is only doing what is expected, so this can't be reduced. However, software does not always use resources optimally and software updates can significantly improve or degrade performance and resource consumption.¶
This exists for assessing power efficiency of Web application components with good results. As an example, GreenIT is available as a browser plugin https://github.com/cnumr/GreenIT-Analysis.¶
Currently network devices are mostly always-on. The design of their software components do not include dynamic power management.¶
The update of legacy networks and devices to support dynamic power management is a complex topic because it impacts the different type of components:¶
hardware components must support variation of power, of bit rate or only being shutdown and restarted¶
firmware components must expose monitoring and actionable functions to software components¶
software components must be updated or added to operate these new capabilities¶
Legacy hardware components are rarely designed to be stopped and restarted. The rhythm of start/stop supported by such components must be documented to prevent wrong usage of their real capacity. This must be present in the datasheet or exposed by the components themselves.¶
In the short-term, as promoted by several discussions in the GREEN-BoF mailing list, the device level assessment requires datamodels augmentation [IVY-WG] to expose these capabilities and configuration updates first, and metrics to measure the power consumption [POWEFF] second.¶
In parallel it should document the minimum information (readable and parsable) about the parts of a legacy device which migrated to sustainability.¶
Hackathon sessions can be used to compare assessment methods based either on inventory metadata or metrics.¶
Legacy network devices energy efficiency monitoring and control differ from end-user devices.¶
As they are alway powered, network devices power consumption is taked in account during the dimensioning of a network. The operational phase, so after deployment, represents the main portion of a network device's total energy consumption. Nevertheless the manufacturing (including testing, ...) and the deployment (including transport...) is significant especially if the recycling is included.¶
On the opposite, end-user portable devices power consumption is explictly managed as they rely on battery and their maximum power consumtion occur during their manufacturing.¶
There are many other types of connected devices, none of these types of devices expose explicitly their energy efficiency.¶
10 years ago the EMAN WG produced the specifications (requirements in rfc6988, framework in rfc7326, ENTITY-MIB in rfc6933, Energy Object Context MIB in rfc7461, Battery Monitoring MIB in rfc7577, Monitoring and Control MIB for Power and Energy in rfc7460, Applicability Statement in rfc7603) for managing the energy produced and consumed by a broad types of network devices. The holistic approach includes standard for devices components rule and identifier.¶
10 years later, there isn't a commonly deployed framework for energy management of network devices components. Despite this, there migh be enough information in EMAN documents to standard the identification of components which "migrated to sustainability" in a legacy device.¶
Energy Object Context MIB [rfc7461], ENTITY-MIB [rfc6933] are reasonable starting points as they provide identification and context definitions which can be reused to describe these components. UUID [rfc9562] is such an example as it is in use both in EMAN MIBs and in YANG Hardware Management [rfc8348].¶
As recalled below in the use case Embedded Carbon, the energy allocated to the manufacturing phase (including the steps before the deployement) is commonly accounted in 'embedded carbon' unit while the power consumption in operation is ususally measured in kilowatt-hour.¶
Legacy hardware had to be replaced after reaching end-of-life (6+ years of constant use), replacement components for older hardware became scarcer.¶
Two objectives of the replacement:¶
reduce power consumption using newer hardware¶
reduce footprint in datacenter using denser devices (i.e., ~120 devices replaced by 30)¶
Before the project started it was difficult to estimate the power consumption reduction of the project. Only some components had idle consumption and max consumption power metrics, which would be used to compare inventory before and after.¶
During and after the project, active power consumption could only be measured from some components (e.g., cpu power consumption estimated via current and voltage sensors), and from the power distribution at each rack (where only a subset of devices were replaced).¶
One goal of the project could not be achieved: measure impact of increased density of devices (i.e., fan size impact moving from 1U to 2U devices which can be attributed to cooling) due to insufficient granularity of power consumption metrics (at component level).¶
Standardization of power consumption metrics (e.g. idle/max) for inventory would improve measurement for legacy hardware replacement projects. Improved power delivery designs inside racks already improves ability to measure real-time device power consumption, this allows device replacement projects to accurately measure impact. Devices reporting granular per-component power consumption metrics would allow component replacement projects to accurately measure reduction.¶
In the pursuit to asses the network is essential to understand the 'embedded carbon' of ICT equipment, which represents the greenhouse gas emissions associated with a product's entire lifecycle. A Life Cycle Assessment (LCA), conforming to international standards such as ISO 14040 and ISO 14044, is employed to model these impacts across multiple categories, from cradle to grave, please refer to Sustainability Insights draft [SUST-INS].¶
For smartphones, approximately 80% of lifetime CO2e emissions are attributed to the production phase, underscoring the significance of manufacturing in the device's embedded carbon. In contrast, networking equipment not only incurs a substantial portion of its embedded carbon during production and testing but also during its operational life, including periods of idle power.¶
Vendors are increasingly focusing on reducing the energy consumption of networking equipment, both in operation and throughout the production process. Modular design and a circular closed-loop approach for component return, repair, recovery, and reuse are instrumental in decreasing the raw materials and emissions required for new components, thereby reducing the embedded carbon.¶
Embedded carbon is typically estimated based on energy consumption at each lifecycle stage, as direct measurement is often impractical. This estimation is crucial for organizations and consumers making environmentally responsible choices and for manufacturers striving to minimize the carbon footprint of their products. By selecting materials and processes with lower associated GHG emissions and improving design, manufacturers can significantly reduce energy consumption and, consequently, embedded carbon.¶
When accounting for the energy consumption of network equipment and components, it is important to consider not only the idle power but also the energy allocated to the manufacturing phase. This comprehensive approach ensures that the embedded energy (and by extension, the embedded carbon) of each piece of equipment is accurately reflected in environmental impact assessments.¶
The tracking of LCM information may reveal device usage information.¶
Device inventory can be used by attackers when assessing a system for vulnerable components.¶
At this step this document has no IANA actions.¶
TODO acknowledge.¶