Energy Metrics For Data Networks
draft-bogdanovic-green-energy-metrics-00

Network Working Group                                      D. Bogdanovic
Internet-Draft                                                     T. Li
Intended status: Informational                          Juniper Networks
Expires: 17 April 2025                                   14 October 2024

                    Energy Metrics For Data Networks
                draft-bogdanovic-green-energy-metrics-00

Abstract

   This document defines a set of energy efficiency metrics to assess
   and optimize the energy consumption of data networks.  These metrics
   enable network administrators and designers to identify opportunities
   for energy savings, optimize network performance, and reduce the
   environmental impact of network operations.  The proposed metrics
   Power Consumption per Data Rate (PCDR), Power Usage Effectiveness
   (PUE), Network Equipment Energy Efficiency (NEEE), and Energy
   Proportionality Coefficient (EPC).

Status of This Memo

   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 17 April 2025.

Copyright Notice

   Copyright (c) 2024 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

Bogdanovic & Li           Expires 17 April 2025                 [Page 1]
Internet-Draft               energy metrics                 October 2024

   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.

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   2
   2.  Standard Testing Conditions . . . . . . . . . . . . . . . . .   2
   3.  Energy efficiency metrics . . . . . . . . . . . . . . . . . .   4
     3.1.  Power Consumption per Data Rate (PCDR)  . . . . . . . . .   4
     3.2.  Power Usage Effectiveness (PUE) . . . . . . . . . . . . .   5
     3.3.  Network Equipment Energy Efficiency (NEEE)  . . . . . . .   6
     3.4.  Energy Proportionality Coefficient (EPC)  . . . . . . . .   7
   4.  Security Considerations . . . . . . . . . . . . . . . . . . .   8
   5.  IANA Considerations . . . . . . . . . . . . . . . . . . . . .   8
   6.  Acknowledgements  . . . . . . . . . . . . . . . . . . . . . .   8
   7.  Change log [RFC Editor: Please remove]  . . . . . . . . . . .   8
   8.  References  . . . . . . . . . . . . . . . . . . . . . . . . .   8
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .   8

1.  Introduction

   As the demand for data network services continues to grow, so does
   the energy consumption associated with operating these networks.
   Improving energy efficiency is crucial for reducing operational costs
   and minimizing environmental impact.  This document defines key
   metrics for evaluating the energy efficiency of data networks,
   providing a comprehensive understanding of energy consumption across
   different network components and layers.

2.  Standard Testing Conditions

   Standardized testing conditions (STC) are crucial for achieving
   reliable and consistent results.  They are essential for:

   *  Ensure Fairness and Objectivity

   *  Enable Valid Comparisons

   *  Supporting Test Validity and Reliability

   *  Minimizing Confounding Variables

Bogdanovic & Li           Expires 17 April 2025                 [Page 2]
Internet-Draft               energy metrics                 October 2024

   While it's impossible to eliminate all external influences,
   standardized testing conditions help reduce their impact on test
   results.  By adhering to standardized testing conditions, companies
   can ensure accurate measurements, fair comparisons, reliable results,
   consistent quality, and compliance with industry standards.  In this
   draft we will focus on two testing conditions:

   a.  Measurement period

   Refers to total duration over which measurements are collected and
   averaged

   1.  24-hour period - it captures a full day-night cycle of network
       usage.  This period accounts for:

       *  Peak and off-peak hours

       *  Variations in network traffic patterns

       *  Daily maintenance windows or scheduled tasks

   2.  Week-long period - provides a more comprehensive view, capturing:

       *  Workday vs. weekend patterns

       *  Weekly maintenance schedules

       *  Longer-term trends

   3.  Short term intervals, e.g. 5 - 15 min, within longer periods
       allow for:

       *  Identification of short-term spikes in power consumption or
          data rates

       *  More accurate averaging of PCDR over time

       *  Correlation with specific network events or applications

       Short term interval should be the same as the bandwidth
       measurement interval at the particular network operator

   b.  Environmental factors

Bogdanovic & Li           Expires 17 April 2025                 [Page 3]
Internet-Draft               energy metrics                 October 2024

   Several organizations define standardized testing conditions in the
   technology sector, mostly on Standard temperature and pressure (STP),
   defined by the National Institute of Standards and Technology (NIST)
   and the International Organization for Standardization (ISO).  There
   are two STP variants to highlight, which are very similar except for
   differences in humidity:

   1.  [ISO13443]

       *  Temperature: 15C (59F)

       *  Pressure: 101.325 kPas (760 mm Hg)

       *  Humidity: 0%

   2.  [NIST]

       *  Temperature: 15C (59F)

       *  Pressure: 101.325 kPas (760 mm Hg)

       *  Humidity: measured during the testing

   For some testing procedures, we will require that the testing
   conditions be measured and reported in the results.  For others,
   adherence to either the ISO or the NIST/EPA testing conditions will
   be required.

3.  Energy efficiency metrics

3.1.  Power Consumption per Data Rate (PCDR)

   Power Consumption per Data Rate (PCDR) measures the amount of power
   consumed to transmit data at a specific rate.  It is expressed in
   watts per gigabit per second (W/(Gbps)).  Using watts as the unit
   provides a clear understanding of how much power is required to
   sustain a certain data transmission rate, which is crucial for
   evaluating the efficiency of network devices and links.

   The PCDR metric applies explicitly to a single link (connection
   between two neighboring nodes at the same layer in the network) or
   group of links or to the whole network as a system.

   PCDR is calculated based on current, actual performance, not
   potential or theoretical capacity.  It uses the actual power consumed
   and the actual data transmission rate during a defined measurement
   period, reflecting real-world operating conditions for a fair basis
   of comparison.

Bogdanovic & Li           Expires 17 April 2025                 [Page 4]
Internet-Draft               energy metrics                 October 2024

   Environmental factors such as ambient temperature, altitude, and
   humidity significantly affect power consumption and, consequently,
   the PCDR.  The testing should follow the ISO 13443 or NIST/EPA
   Standard.  If setting those environmental variables is not possible,
   the environmental conditions during testing must must be measured and
   documented to ensure transparency:

   Ambient Temperature: C (Centigrade)
   Pressure: atm (atmoshere)
   relative humidity: percentage

 Formula:
           Power Consumption (W)
 PCDR = -----------------------------
          Transmission Rate (Gbps)

 Where:
 Power Consumption (W): The average power in watts consumed by the
 specified link, group of links or the network during the measurement
 period.

 Data Transmission Rate (Gbps): The average data rate in bits per second
 transmitted over the link, group of links or the network during the
 same period.

3.2.  Power Usage Effectiveness (PUE)

   Power Usage Effectiveness (PUE), originally used in data centers, and
   can be reused in the networking space.  PUE is calculated as the
   ratio of the energy consumed by the actual networking equipment
   (switches, routers, etc.) to total energy consumed by the network
   supporting infrastructure (building, cooling, etc.  This metric helps
   understand the efficiency of the network and supporting
   infrastructure.

   Formula:
            Energy Consumed by Networking Equipment (kWh)
   PUE = -----------------------------------------------------
         Total Energy Consumed by Network Infrastructure (kWh)

   A good PUE is typically considered to be around 1.5 or lower.

   1.0 to 1.2: This range is considered excellent.  It means that almost
   all the energy consumed by the network is used for data transmission
   (network equipment), with very little energy wasted on cooling,
   lighting, and other non-IT systems.  Achieving this level typically
   requires efficient energy management practices.

Bogdanovic & Li           Expires 17 April 2025                 [Page 5]
Internet-Draft               energy metrics                 October 2024

   1.2 to 1.5: This is still very good and indicates a highly efficient
   network.

   Above 1.5 suggests that the network has room for improvement in its
   energy efficiency.  A higher PUE means inefficient network equipment
   (older generation) and there is lot of room for improvement.

   The goal is to get as close to 1.0 as possible, indicating that
   nearly all the energy is being used directly for network operations.
   However, achieving extremely low PUE values can be challenging and
   may involve substantial investments in energy-efficient
   infrastructure.

3.3.  Network Equipment Energy Efficiency (NEEE)

   Network Equipment Energy Efficiency (NEEE) measures the energy
   efficiency of network devices such as switches, routers, etc.  It is
   calculated as the ratio of data throughput (bits per second) to the
   power consumption (watts), expressed in bits per second per watt
   (bps/W).  This metric provides insight into how efficiently a network
   device can forward data relative to the power it consumes.

   Data Throughput (Gbps): The actual, measured data rate at which the
   network device is forwarding data during the measurement period.
   It's essential to specify the exact bandwidth involved because NEEE
   can vary significantly with different throughput levels.  For
   meaningful and comparable results, measurements should be taken at
   defined bandwidth levels that reflect typical operational scenarios,
   such as:

   Idle State (0 bps): When the device is powered on but not forwarding
   any data.

   Low Utilization: A percentage (15%) of the device's maximum
   forwarding capacity.

   Medium Utilization: 50% of maximum capacity.

   High Utilization: Near the device's maximum forwarding capacity
   (90%).

   Idle Power Consumption: Even when not actively forwarding data,
   network devices consume a baseline amount of power (idle power) to
   stay operational.  This idle power consumption contributes to the
   overall energy usage and should be considered when evaluating energy
   efficiency.

   Importance of Measuring at Specific Bandwidth Levels:

Bogdanovic & Li           Expires 17 April 2025                 [Page 6]
Internet-Draft               energy metrics                 October 2024

   Load-Dependent Efficiency: NEEE can vary depending on the network
   device's load.  Devices may be more efficient at certain throughput
   levels due to factors like hardware design and power scaling
   features.

   For fair comparisons, NEEE should be measured under standardized
   conditions, as described in Standard Testing Conditions (STC),
   including specified bandwidth levels.

   Formula:

           Data Throughput (Gbps)
   NEEE = -----------------------
           Power Consumption (W)

3.4.  Energy Proportionality Coefficient (EPC)

   Energy proportionality coefficient is significant in energy
   efficiency metrics because network devices rarely operate at maximum
   capacity continuously, so metrics must account for varying levels of
   utilization to provide a realistic assessment.  By incorporating
   energy proportionality into metrics like the Network Equipment Energy
   Efficiency (NEEE), evaluations can reflect a device's performance
   across different operational states, from idle to peak throughput.
   Since devices consume a baseline amount of power even when not
   actively processing data, energy proportionality aims to minimize
   this idle consumption.  Designing devices that consume significantly
   less power at low loads improves overall energy efficiency, which is
   especially beneficial in networks with fluctuating traffic patterns.

   Energy proportionality offers significant benefits, including reduced
   energy consumption and improved device efficiency.  By lowering power
   consumption during periods of low network activity, organizations
   achieve operational savings through lower energy bills and contribute
   to environmental sustainability by decreasing their carbon footprint
   and aligning with sustainability goals.  Additionally, devices
   designed with energy proportionality in mind operate efficiently
   across all workloads, ensuring optimal performance.  Reduced thermal
   stress from lower power usage can also lead to an extended hardware
   lifespan, enhancing the longevity and reliability of the equipment.

   Definition: A metric that quantifies how closely a device's power
   consumption scales with its workload.

Bogdanovic & Li           Expires 17 April 2025                 [Page 7]
Internet-Draft               energy metrics                 October 2024

   Calculation:

           Power Consumption at Given Load − Idle Power Consumption
   EPC = ------------------------------------------------------------
           Maximum Power Consumption − Idle Power Consumption

   Interpretation: An EPC value closer to 1 indicates better energy
   proportionality at that specific load level.

   Standardized Testing Conditions: Ensure consistent environmental
   factors per STC during testing.

   Workload Profiles: Use realistic traffic patterns to simulate
   different network conditions using consistent device physical (number
   of physical links inserted) and logical configuration.

   Data Collection: Record power consumption and throughput at each
   workload level to compute NEEE and EPC.

4.  Security Considerations

   TBD

5.  IANA Considerations

   This document currently has no items for IANA considerations.

6.  Acknowledgements

7.  Change log [RFC Editor: Please remove]

8.  References

   [ISO13443] ISO, "ISO 13443 Natural gas — Standard reference
              conditions", <https://www.iso.org/obp/
              ui/#iso:std:iso:13443:ed-1:v1:en>.

   [NIST]     NIST, "Standard Temperature and Pressure",
              <https://en.wikipedia.org/wiki/
              Standard_temperature_and_pressure#cite_ref-21>.

Authors' Addresses

   Dean Bogdanovic
   Juniper Networks
   Email: dean.b@juniper.net

Bogdanovic & Li           Expires 17 April 2025                 [Page 8]
Internet-Draft               energy metrics                 October 2024

   Tony Li
   Juniper Networks
   Email: tli@juniper.net

Bogdanovic & Li           Expires 17 April 2025                 [Page 9]