Network Working Group B. Claise
Internet-Draft J. Parello
Intended Status: Informational B. Schoening
Expires: April 20, 2011 Cisco Systems, Inc.
October 20, 2010
Power Management Architecture
draft-claise-power-management-arch-02
Status of this Memo
This Internet-Draft is submitted to IETF in full conformance
with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet
Engineering Task Force (IETF), its areas, and its working
groups. Note that other groups may also distribute working
documents as Internet-Drafts.
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."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed
at http://www.ietf.org/shadow.html
This Internet-Draft will expire on September, 2010.
<Claise, et. Al> Expires April 20 2011 [Page 1]
Internet-Draft <Power Management Archictecure> Octobre 2010
Copyright Notice
Copyright (c) 2010 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
(http://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 Simplified BSD License text as described
in Section 4.e of the Trust Legal Provisions and are provided
without warranty as described in the Simplified BSD License.
Abstract
This document defines the power management architecture.
<Claise, et. Al> Expires April 20, 2011 [Page 2]
Internet-Draft <Power Management Archictecure> Octobre 2010
Table of Contents
1. Introduction.............................................. 4
2. Use Cases & Requirements.................................. 5
3. Terminology............................................... 5
4. Energy Management Reference Model......................... 7
5. Architecture High Level Concepts and Scope................ 9
5.1. Power Monitor Information........................... 11
5.2. Power Monitor Meter Domain.......................... 11
5.3. Power Monitor Parent and Child...................... 11
5.4. Power Monitor Context............................... 12
5.5. Power Monitor Levels................................ 13
5.6. Power Monitor Usage Measurement..................... 16
5.7. Optional Power Usage Quality........................ 17
5.8. Optional Energy Measurement......................... 18
5.9. Optional Battery Information........................ 18
6. Power Monitor Children Discovery......................... 18
7. Configuration............................................ 19
8. Fault Management......................................... 20
9. IPFIX.................................................... 20
10. Relationship with Other Standards Development
Organizations............................................... 21
10.1. Information Modeling............................... 21
10.2. Power Levels....................................... 21
11. Implementation Scenarios................................ 22
Scenario 1: Switch with PoE endpoints.................... 22
Scenario 2: Switch with PoE endpoints with further connected
device(s)................................................ 22
Scenario 3: A switch with Wireless Access Points......... 22
Scenario 4: Network connected facilities gateway......... 23
Scenario 5: Data center network.......................... 23
Scenario 6: Building gateway device...................... 23
Scenario 7: Power consumption of UPS..................... 23
Scenario 8: Power consumption of battery-based devices... 24
12. Security Considerations................................. 24
12.1. Security Considerations for SNMP...................... 24
12.2. Security Considerations for IPFIX..................... 25
13. IANA Considerations..................................... 25
14. Acknowledgments......................................... 25
15. References.............................................. 25
Normative References..................................... 25
Informative References................................... 26
<Claise, et. Al> Expires April 20, 2011 [Page 3]
Internet-Draft <Power Management Archictecure> Octobre 2010
TO DO
- Question for the Working Group: Should the WG consider
IPFIX in this architecture?
- Question for the Working Group: How to specify the notion
of child capabilities, i.e. the capabilities that the
Power Monitor Parents have with Power Monitor Children.
For Example:
1. Monitoring (only reporting)
2. Configuration power state
3. Configuration: power
Example: on a PC, we can set power level without knowing
the power. A solution must be specified in this draft.
- Question for the Working Group: Should transition states
be tracked when setting a level. Example: The configured
level is set to Off from High. The Actual level will
take time to update as the device powers down. Should
there be transitions shown or will the two variables
suffice to track the device state.
- Question for working group: Should implementation
scenarios be incorporated in the architecture draft
- We should have a similar section, for all the drafts,
which includes an overview of all EMAN documents.
1. Introduction
Network management is typically divided into the five main
network management areas defined in the ISO Telecommunications
Management Network model: Fault, Configuration, Accounting,
Performance, and Security Management. Absent from this model is
any consideration of energy management, which is now becoming a
critical area of concern worldwide.
This document defines an architecture for power management for
devices within or connected to communication networks. This
architecture includes monitoring for power state and energy
consumption of networked elements, covering the requirements
specified in [POWER-MON-REQ]. It also goes a step further in
defining some elements of configuration.
<Claise, et. Al> Expires April 20, 2011 [Page 4]
Internet-Draft <Power Management Archictecure> Octobre 2010
Energy management is applicable to devices in communication
networks. Target devices for this specification include (but
are not limited to): routers, switches, Power over Ethernet
(PoE) endpoints, protocol gateways for building management
systems, intelligent meters, home energy gateway, hosts and
servers, sensor proxies, etc.
Where applicable, device monitoring extends to the individual
components of the device and to any attached dependent devices.
For example: A device can contain components that are
independent from a power-state point of view, such as line
cards, processor cards, hard drives. A device can also have
dependent attached devices, such as a switch with PoE endpoints
or a power distribution unit with attached endpoints.
2. Use Cases & Requirements
Requirements for power and energy monitoring for networking
devices are specified in [POWER-MON-REQ]. The requirements in
[POWER-MON-REQ] cover devices typically found in communications
networks, such as switches, routers, and various connected
endpoints. For a power monitoring architecture to be useful, it
should also apply to facility meters, power distribution units,
gateway proxies for commercial building control, home automation
devices, and devices that interface with the utility and/or
smart grid. Accordingly, this architecture, the scope is
broader than that specified in [POWER-MON-REQ]. Several
scenarios that cover these broader use cases are presented later
in Section 11. - Implementation Scenarios.
3. Terminology
This section contains definitions of important terms used
throughout this specification.
IPFIX-specific terminology used in this document is defined in
section 2 of [RFC5101]. For example: Flow Record, Collector ,
etc... As in [RFC5101], these IPFIX-specific terms have the
first letter of a word capitalized.
Power Monitor
A Power Monitor is a component within a system of components
that provides power, draws power, or reports energy consumption
on behalf of another Power Monitor. It can be independently
<Claise, et. Al> Expires April 20, 2011 [Page 5]
Internet-Draft <Power Management Archictecure> Octobre 2010
managed from a power-monitoring and power-state configuration
point of view. Examples of Power Monitors are: a router line
card, a motherboard with a CPU, an IP phone connected with a
switch, etc.
Power Monitor Parent
A Power Monitor Parent is a Power Monitor that is the root of
one or more subtending Power Monitors, called Power Monitor
Children. The Power Monitor Parent is able to collect data
about or report on the power state and energy consumption of its
Power Monitor Children.
For example: A Power-over-Ethernet (PoE) device (such as an IP
phone or an access point) is attached to a switch port. The
switch is the source of power for the attached device, so the
Power Monitor Parent is the switch, and the Power Monitor Child
is the device attached to the switch.
The Power Monitor Parent may report data or implement actions on
behalf of the Power Monitor Child. These capabilities must be
enumerated by the Power Monitor Parent.
The communication between the parent and child for monitoring or
collection of power data is left to the device manufacturer. For
example: A parent switch may use LLDP to communicate with a
connected child, and a parent lighting controller may use BACNET
to communicate with child lighting devices.
Power Monitor Child
A Power Monitor Child is a Power Monitor associated with a Power
Monitor Parent, and which reports its power usage and power
state to its Power Monitor Parent. The Power Monitor Child may
or may not draw power from its Power Monitor Parent. .
Power Monitor Meter Domain
A Power Monitor Meter Domain is a name or name space that
logically groups Power Monitors into a zone of manageable power
usage. Typically, this zone will have as members all Power
Monitors that are powered from the same electrical panel or
panels for which there is a meter or sub meter. For example:
All Power Monitors receiving power from the same distribution
panel of a building, or all Power Monitors in a building for
<Claise, et. Al> Expires April 20, 2011 [Page 6]
Internet-Draft <Power Management Archictecure> Octobre 2010
which there is one main meter, would comprise a Power Monitor
Meter Doman. From the standpoint of power-use monitoring, it is
useful to report the total power usage as the sum of power
consumed by all the Power Monitors within a Power Monitor Meter
Domain and then correlate that value with the metered usage.
Power Level
A Power Level is a uniform way to classify power settings on a
Power Monitor (e.g., shut, hibernate, sleep, high). Power
Levels can be viewed as an interface for the underlying device-
implemented power settings.
Manufacturer Power Level
A Manufacturer Power Level is a device-specific way to classify
power settings implemented on a Power Monitor. For cases where
the implemented power settings cannot be directly mapped to
Power Levels, we can use the Manufacturer Power Levels to
enumerate and show the relationship between the implemented
power settings and the Power Level interface.
4. Energy Management Reference Model
+---------------+
| NMS | -
+-----+---+-----+ |
| | |
| | | S
+---------+ +-------+ | N
| | | M
| | | P
+---------------+ +------+-------+ |
| Power Monitor | | Power Monitor | |
| Parent 1 | ... | Parent N | -
+---------------+ +---------------+
|||
(protocol |||
out of ||| +-------------+---------+
the scope)|||------| Power Monitor Child 1 |
|| +-----------------------+
||
|| +-------------+---------+
||-------| Power Monitor Child 2 |
<Claise, et. Al> Expires April 20, 2011 [Page 7]
Internet-Draft <Power Management Archictecure> Octobre 2010
| +-----------------------+
|
|
|-------- ...
|
|
| +-------------+---------+
|--------| Power Monitor Child M |
+-----------------------+
Figure 1: Energy Management Pull Reference Model
In this architecture a Network Management Station (NMS) will
poll MIB variables on a Power Monitors via SNMP. The Power
Monitor returns information for itself and for any Power Monitor
Children if applicable. The information returned will contain
business context, energy usage, power quality and other
information as described further.
The protocol between the Power Monitor Parent and Power Monitor
Children is out of scope of this document. The Power Monitor
Parent may speak to a Power Monitor Child using a manufacturer
selected protocol. This protocol may or may not based on IP.
In this way, a Power Monitor Parent acts as a PROXY for protocol
translation between the Power Monitor Parent and Child. The
Power Monitor Parent also acts as an aggregation point for other
subtended Power Monitor Children.
+---------------+
| NMS/Collector | ^ S
+-----+---+-----+ | N
| | | M I
| | | P P
+---------+ +-------+ | & F
| | | N I
| | | O X
+---------------+ +------+-------+ | T
| Power Monitor | | Power Monitor | | .
| Parent 1 | ... | Parent N | -
+---------------+ +---------------+
|||
(protocol |||
out of ||| +-------------+---------+
the scope)|||------| Power Monitor Child 1 |
|| +-----------------------+
||
<Claise, et. Al> Expires April 20, 2011 [Page 8]
Internet-Draft <Power Management Archictecure> Octobre 2010
|| +-------------+---------+
||-------| Power Monitor Child 2 |
| +-----------------------+
|
|
|-------- ...
|
|
| +-------------+---------+
|--------| Power Monitor Child M |
+-----------------------+
Figure 2: Energy Management PUSH Reference Model
The Power Monitor Parents may send SNMP notifications regarding
their own state or the state of their Power Monitor Children.
The Power Monitor Children do not send SNMP notifications on
their own.
As discussed in [POWER-MON-REQ], the Power Monitor Parents may
export IPFIX Flow Records [RFC5101] to a Collector. The IPFIX
protocol is well suited for regular time series export of
similar information, such as the energy consumed by the Power
Monitor Children.
EDITOR'S NOTE: at this point in time, there is no draft
specifying the IPFIX Flow Records.
5. Architecture High Level Concepts and Scope
The scope of this architecture is to enable networking and
network-attached devices to be managed with respect to their
energy consumption or production. The goal is to make devices
energy-aware.
The architecture describes how to make a device aware of its
consumption or production of energy expressed as usage in watts.
This does not include:
- Manufacturing costs in currency or environmental units
- Embedded carbon or environmental equivalences of the device
itself
- Cost in currency or environmental impact to dismantle or
recycle the device
- Relationship to an electrical or smart grid
- Supply chain analysis
<Claise, et. Al> Expires April 20, 2011 [Page 9]
Internet-Draft <Power Management Archictecure> Octobre 2010
- Conversion of the usage or production of energy to units
expressed from the source of that energy (such as the greenhouse
gas emissions associated with 1000kW from a diesel source).
The remainder of this section describes the basic concepts of
the architecture. Each concept is examined in detail in
subsequent sections.
Examples are provided in a later section to show how these
concepts can be implemented.
The basic concepts are:
The Power Monitor will have basic naming and informational
descriptors to identify it in the network.
A Power Monitor can be part of a Power Monitor Meter Domain. A
Power Monitor Meter Domain is a manageable set of devices that
has a meter or sub-meter attached and typically corresponds to a
power distribution point or panel.
A Power Monitor can be a parent (Power Monitor Parent) or child
(Power Monitor Child) of another Power Monitor. This allows for
Power Monitor Parent to aggregate power reporting and control of
power information.
Each Power Monitor can have information to allow it to be
described in the context of the business or ultimate use. This
is in addition to its networked information. This allows for
tagging, grouping, and differentiation between Power Monitors
for NMS.
For control and universal monitoring, each Power Monitor
implements or declares a set of known Power Levels. The Power
Levels are mapped to Manufacturer Power Levels that indicate the
specific power settings for the device implementing the Power
Monitor.
When the Power Level is set, a Power Monitor may be busy at the
request time. The Power Monitor will set the desired level and
then update the actual Power Level when the priority task is
finished. This mechanism implies two different Power Level
variables: actual versus desired.
EDITOR'S NOTE: The transition state will have to be specified.
<Claise, et. Al> Expires April 20, 2011 [Page 10]
Internet-Draft <Power Management Archictecure> Octobre 2010
Each Power Monitor will have usage information that describes
the power information along with how that usage was obtained or
derived.
Optionally, a Power Monitor can further describe the power
information with power quality information reflecting the
electrical characteristics of the measurement.
Optionally, a Power Monitor can provide power usage over time to
describe energy consumption
If a Power Monitor has one or more batteries, it can provide
optional battery information as well.
5.1. Power Monitor Information
Every Power Monitor should have a unique printable name, and
must have a unique Power Monitor index.
Possible naming conventions are: textual DNS name, MAC-address
of the device, interface ifName, or a text string uniquely
identifying the Power Monitor. As an example, in the case of IP
phones, the Power Monitor name can be the device DNS name.
5.2. Power Monitor Meter Domain
Each Power Monitor must be a member of a Power Monitor Meter
Domain. The Power Monitor Meter Domain should map 1-1 with a
metered or sub-metered portion of the site. The Power Monitor
Meter Domain must be configured on the Power Monitor Parent.
The Power Monitor Children may inherit their domain values from
the Power Monitor Parent or the Power Monitor Meter Domain may
be configured directly in a Power Monitor Child.
5.3. Power Monitor Parent and Child
A Power Monitor Child reports its power usage to its Power
Monitor Parent. A Power Monitor Child has one and only one
Power Monitor Parent. If a Power Monitor had two parents there
would be a risk of double-reporting the power usage in the Power
Monitor Meter Domain. Therefore, a Power Monitor cannot be both
a Power Monitor Parent and a Power Monitor Child at the same
time.
<Claise, et. Al> Expires April 20, 2011 [Page 11]
Internet-Draft <Power Management Archictecure> Octobre 2010
A Power Monitor Child can be fully dependent on the Power
Monitor Parent for its power or independent from the parent
(such as a PC connected to a switch). In the dependently
powered case, the Power Monitor Parent provides power for the
Power Monitor Child (as in the case of Power Over Ethernet
devices). In the independently powered case, the Power Monitor
Child draws power from another source (typically a wall outlet).
Since the Power Monitor Parent is not the source of power
supply, the power usage cannot be measured at the Power Monitor
Parent. However, an independent Power Monitor Child reports
Power Monitor information to the Power Monitor Parent. The
Power Monitor Child may listen to the power control settings
from a Power Monitor Parent and could react to the control
messages. However, note that the communication between the
Power Monitor Parent and Power Monitor Child is out of scope for
this document.
A mechanism, outside of the scope of this document, should be in
place to verify the connectivity between the Power Monitor
Parent and its Power Monitor Children. If a Power Monitor Child
is unavailable, the Power Monitor Parent must follow some rules
to determine how long it should wait before removing the Power
Monitor Child entry, along with all associated statistics, from
its database. In some situations, such as a connected building
in which the Power Monitor Children are somewhat static, this
removal-delay period may be long, and persistence across a Power
Monitor Parent reload may make sense. However, in a networking
environment, where endpoints can come and go, there is not much
sense in configuring a long removal timer. In all cases, the
removal timer or persistence must be clearly specified.
Further examples of Power Monitor Parent and Child
implementations are provided in the Implementation Scenarios
section 11.
5.4. Power Monitor Context
Monitored power data will ultimately be collected by and
reported from an NMS. In order to aid in reporting and in
differentiation between Power Monitors, each Power Monitor will
contain information establishing its business or site context.
A Power Monitor can provide an importance value in the range of
1 to 100 to help differentiate a device's use or relative value
to the site. The importance range is from 1 (least important)
to 100 (most important). The default importance value is 1.
<Claise, et. Al> Expires April 20, 2011 [Page 12]
Internet-Draft <Power Management Archictecure> Octobre 2010
For example: A typical office environment has several types of
phones, which can be rated according to their business impact.
A public desk phone has a lower importance (for example, 10)
than a business-critical emergency phone (for example, 100). As
another example: A company can consider that a PC and a phone
for a customer-service engineer is more important than a PC and
a phone for lobby use.
Although network managers must establish their own ranking, the
following is a broad recommendation:
. 90 to 100 Emergency response
. 80 to 90 Executive or business-critical
. 70 to 79 General or Average
. 60 to 69 Staff or support
. 40 to 59 Public or guest
. 1 to 39 Decorative or hospitality
A Power Monitor can provide a set of keywords. These keywords
are a list of tags that can be used for grouping and summary
reporting within or between Power Monitor Meter Domains. All
alphanumeric characters and symbols, such as #, (, $, !, and &,
are allowed. Potential examples are: IT, lobby, HumanResources,
Accounting, StoreRoom, CustomerSpace, router, phone, floor2, or
SoftwareLab. There is no default value for a keyword.
Multiple keywords can be assigned to a device. In such cases,
the keywords are separated by commas and no spaces between
keywords are allowed. For example, "HR,Bldg1,Private".
Additionally, a Power Monitor can provide a "role description"
string that indicates the purpose the Power Monitor serves in
the network or for the site/business. This could be a string
describing the context the device fulfills in deployment. For
example, a lighting fixture in a kitchen area could have a role
of "Hospitality Lighting" to provide context for the use of the
device.
5.5. Power Monitor Levels
Power Levels represent universal states of power management of a
Power Monitor. Each Power Level corresponds to a global,
system, and performance state in the ACPI model [ACPI].
Level ACPI Global/System Power Level
State Name
<Claise, et. Al> Expires April 20, 2011 [Page 13]
Internet-Draft <Power Management Archictecure> Octobre 2010
Non-operational states:
1 G3, S5 Mech Off
2 G2, S5 Soft Off
3 G1, S4 Hibernate
4 G1, S3 Sleep
5 G1, S2 Standby
6 G1, S1 Ready
Operational states:
7 G0, S0, P5 LowMinus
8 G0, S0, P4 Low
9 G0, S0, P3 MediumMinus
10 G0, S0, P2 Medium
11 G0, S0, P1 HighMinus
12 G0, S0, P0 High
Figure 3: ACPI / Power Level Mapping
For example, a Power Monitor with a Power Level of 9 would
indicate an operational state with MediumMinus Power Level.
The Power Levels can be considered as guidelines in order to
promote interoperability across device types. Realistically,
each specific feature requiring Power Levels will require a
complete recommendation of its own. For example, designing IP
phones with consistent Power Levels across vendors requires a
specification for IP phone design, along with the Power Levels
mapping.
Manufacturer Power Levels are required in some situations, such
as when no mappings with the existing Power Levels are possible,
or when more than the twelve specified Power Levels are
required.
A first example would be an imaginary device type, with only
five levels: "none", "short", "tall", "grande", and "venti".
Manufacturer Power Level Respective Name
0 none
1 short
2 tall
3 grande
4 venti
Figure 4: Mapping Example 1
<Claise, et. Al> Expires April 20, 2011 [Page 14]
Internet-Draft <Power Management Archictecure> Octobre 2010
In the unlikely event that there is no possible mapping between
these Manufacturer Power Levels and the proposed Power Monitor
Power Levels, the Power Level will remain 0 throughout the MIB
module, as displayed below.
Power Level / Name Manufacturer Power Level / Name
0 / unknown 0 / none
0 / unknown 1 / short
0 / unknown 2 / tall
0 / unknown 3 / grande
0 / unknown 4 / venti
Figure 5: Mapping Example 2
If a mapping between the Manufacturer Power Levels and the Power
Monitor Power Levels is achievable, both series of levels must
exist in the MIB module in the Power Monitor Parent, allowing
the NMS to understand the mapping between them by correlating
the Power Level with the Manufacturer Power Levels.
Power Level / Name Manufacturer Power Level / Name
1 / Mech Off 0 / none
2 / Soft Off 0 / none
3 / Hibernate 0 / none
4 / Sleep, Save-to-RAM 0 / none
5 / Standby 0 / none
6 / Ready 1 / short
7 / LowMinus 1 / short
8 / Low 1 / short
9 / MediumMinus 2 / tall
10 / Medium 2 / tall
11 / HighMinus 3 / grande
12 / High 4 / venti
Figure 6: Mapping Example 3
How the Power Monitor Levels are then mapped is an
implementation choice. However, it is recommended that the
Manufacturer Power Levels map to the lowest applicable Power
Levels, so that setting all Power Monitors to a Power Level
would be conservative in terms of disabled functionality on the
Power Monitor.
A second example would be a device type, such as a dimmer or a
motor, with a high number of operational levels. For the sake
of the example, 100 operational states are assumed.
<Claise, et. Al> Expires April 20, 2011 [Page 15]
Internet-Draft <Power Management Archictecure> Octobre 2010
Power Level / Name Manufacturer Power Level / Name
1 / Mech Off 0 / off
2 / Soft Off 0 / off
3 / Hibernate 0 / off
4 / Sleep, Save-to-RAM 0 / off
5 / Standby 1 / off
6 / Ready 2 / off
7 / LowMinus 11 / 1%
7 / LowMinus 12 / 2%
7 / LowMinus 13 / 3%
. .
. .
. .
8 / Low 15 / 15%
8 / Low 16 / 16%
8 / Low 17 / 17%
. .
. .
. .
9 / MediumMinus 30 / 30%
9 / MediumMinus 31 / 31%
9 / MediumMinus 32 / 32%
. .
. .
. .
10 / Medium 45 / 45%
10 / Medium 46 / 46%
10 / Medium 47 / 47%
. .
. .
. .
etc...
Figure 7: Mapping Example 4
As specified in section 6, this architecture allows the
configuration of the Power Level, while configuring the
Manufacturer Power Level from the MIB directly is not possible.
5.6. Power Monitor Usage Measurement
The usage or production or power must be qualified as more than
a value alone. A measurement should be qualified with the
units, magnitude, direction of power flow, and by what means the
measurement was made (ex: Root Mean Square versus Nameplate) .
<Claise, et. Al> Expires April 20, 2011 [Page 16]
Internet-Draft <Power Management Archictecure> Octobre 2010
In addition, the Power Monitor should describe how it intends to
measure usage as one of consumer, producer or meter of usage.
Given the intent any readings can be correctly summarized or
analyzed by an NMS. For example metered usage reported by a
meter and consumption usage reported by a device connected to
that meter may naturally measure the same usage. With the two
measurements identified by intent a proper summarization can be
made by an NMS.
The power usage measurement should conform to the IEC 61850
definition of unit multiplier for the SI (System International)
units of measure. The power usage measurement is considered an
instantaneous usage value and does not include the usage over
time.
Measured values are represented in SI units obtained by
BaseValue * 10 raised to the power of the scale. For example,
if current power usage of a Power Monitor is 3, it could be 3 W,
3 mW, 3 KW, or 3 MW, depending on the value of the scaling
factor
In addition to knowing the usage and magnitude, it is useful to
know how a Power Monitor usage measurement was obtained:
. Whether the measurements were made at the device itself or
from a remote source.
. Description of the method that was used to measure the
power and whether this method can distinguish actual or
estimated values.
An NMS can use this information to account for the accuracy and
nature of the reading between different implementations.
In addition to the power usage, the nameplate power rating of a
Power Monitor is typically specified by the vendor as the
capacity required to power the device. Often this label is a
conservative number and is the worst-case power draw. While the
actual utilization of an entity can be lower, the nameplate
power is important for provisioning, capacity planning and
billing.
5.7. Optional Power Usage Quality
Given a power measurement of a Power Monitor, it may in certain
circumstances be desirable to know the power quality associated
with that measurement. The information model must adhere to the
IEC 61850 7-2 standard for describing AC measurements. In some
<Claise, et. Al> Expires April 20, 2011 [Page 17]
Internet-Draft <Power Management Archictecure> Octobre 2010
Power Monitor Domains, the power quality may not be needed,
available, or relevant to the Power Monitor.
5.8. Optional Energy Measurement
In addition to reporting the Power Level, an approach to
characterizing the energy demand is required. It is well known
in commercial electrical utility rates that demand charges can
be on par with actual power charges, so it is useful to
characterize the demand. The demand can be described as the
average energy of an Power Monitor over a time window called a
demand interval (typically 15 minutes). The highest peak energy
demand measured over a time horizon, such as 1 month or 1 year,
is often the basis for usage charges. A single window of time
of high usage can penalize the consumer with higher energy
consumption charges. However, it is relevant to measure the
demand only when there are actual power measurements from a
Power Monitor, and not when the power measurement is assumed or
predicted.
Several efficiency metrics can be derived and tracked with the
demand usage data. For example:
. Per-packet power costs for a networking device (router or
switch) can be calculated by an NMS. The packet count can
be determined from the traffic usage in the ifTable
[RFC2863], from the forwarding plane figure, or from the
platform specifications.
. Watt-hour power can be combined with utility energy sources
to estimate carbon footprint and other emission statistics.
5.9. Optional Battery Information
Some Power Monitors may be running on batteries. Therefore
information such as the battery status (charging or
discharging), remaining capacity, and so on, must be available.
6. Power Monitor Children Discovery
There are multiple ways that the Power Monitor Parent can
discover its Power Monitor Children, if they are not present on
the same physical network element:
<Claise, et. Al> Expires April 20, 2011 [Page 18]
Internet-Draft <Power Management Archictecure> Octobre 2010
. In case of PoE, the Power Monitor Parent automatically
discovers a Power Monitor Child when the Child requests
power.
. The Power Monitor Parent and Children may run the Link
Layer Discovery Protocol [LLDP], or any other discovery
protocol, such as Cisco Discovery Protocol (CDP). The
Power Monitor Parent might even support the LLDP-MED MIB
[LLDP-MED-MIB], which returns extra information on the
Power Monitor Children.
. The Power Monitor Parent may reside on a network connected
facilities gateway. A typical example is a converged
building gateway, monitoring several other devices in the
building, and serving as a proxy between SNMP and a
protocol such as BACNET.
When a Power Monitor Child supports only its own Manufacturer
Power Levels, the Power Monitor Parent will have to discover
those Manufacturer Power Levels. Note that the communication
specifications between the Power Monitor Parent and Children is
out of the scope of this document. This includes the
Manufacturer Power Levels discovery, which is protocol-specific.
7. Configuration
This power management architecture allows the configuration of
the following key parameters:
. Power Monitor name: A unique printable name for the Power
Monitor.
. Power Monitor Role: An administratively assigned name to
indicate the purpose a Power Monitor serves in the network.
. Power Monitor Importance: A ranking of how important the
Power Monitor is, on a scale of 1 to 100, compared with
other Power Monitors in the same Power Monitor Meter
Domain.
. Power Monitor Keywords: A list of keywords that can be used
to group Power Monitors for reporting or searching.
. Power Monitor Domain: Specifies the name of a Power Monitor
Meter Domain for the Power Monitor.
. The Power Monitor Level: Specifies the current Power Level
(0..12) for the Power Monitor.
. The energy demand parameters: For example, which interval
length to report the energy on, the number of intervals to
keep, etc.
When a Power Monitor requires a mapping with the Manufacturer
Power Level, the Power Monitor configuration is done via the
<Claise, et. Al> Expires April 20, 2011 [Page 19]
Internet-Draft <Power Management Archictecure> Octobre 2010
Power Level settings, and not directly via the Manufacturer
Power Levels, which are read-only. Taking into account Figure
8, where the LowMinus Power Level corresponds to three different
Manufacturer Power Levels (11 for 1%, 12 for 2%, and 13 for 3%),
the implication is that this architecture will not set the
Manufacturer Power Level to one percent granularity without
communicating over or configuring the proprietary protocol for
this Power Monitor.
This architecture uses a Power Level MIB object to set up the
Power Level for a specific Power Monitor. However, the Power
Monitor might be busy executing an important task that requires
the current Power Level for some more time. For example, a PC
might have to finish a backup first, or an IP phone might be
busy with a current phone call. Therefore a second MIB object
contains the actual Power Level. A difference in values between
the two objects indicates that the Power Monitor is currently in
Power Level transition.
Interactions with established open protocols, such as Wake-up-
on-Lan (WoL) and DASH [DASH], may require configuration in the
Power Monitor as well, facilitating the communication between
Power Monitor Parent and remote Power Monitor Children.
Note that the communication specifications between the Power
Monitor Parent and Children is out of the scope of this
document. This includes communication of power settings and
configuration information, such as the Power Monitor Domain.
8. Fault Management
[POWER-MON-REQ] specifies some requirements about power states
such as "the current state - the time of the last change", "the
total time spent in each state", "the number of transitions to
each state", etc. Such requirements are fulfilled via the
pmPowerLevelChange NOTIFICATION-TYPE [POWER-MON-MIB]. This SNMP
notification is generated when the value(s) of Power Level has
changed for the Power Monitor.
9. IPFIX
A push-based mechanism, such as IPFIX [RFC5101], might be
required to export high-volume time series of energy consumption
values, as mentioned in [POWER-MON-REQ].
<Claise, et. Al> Expires April 20, 2011 [Page 20]
Internet-Draft <Power Management Archictecure> Octobre 2010
EDITOR'S NOTE: the Working Group should decide how much of IPFIX
should be described in this document
10. Relationship with Other Standards Development Organizations
10.1. Information Modeling
This power management architecture should, as much as possible,
reuse existing standards efforts, especially with respect to
information modeling and data modeling [RFC3444].
The data model for power, energy related objects is based on IEC
61850.
Specific examples include:
. The scaling factor, which represents Power Monitor usage
magnitude, conforms to the IEC 61850 definition of unit
multiplier for the SI (System International) units of
measure.
. The power accuracy model is based on the ANSI and IEC
Standards, which require that we use an accuracy class for
power measurement. ANSI and IEC define the following
accuracy classes for power measurement:
. IEC 62053-22 60044-1 class 0.1, 0.2, 0.5, 1 3.
. ANSI C12.20 class 0.2, 0.5
. The powerQualityMIB MIB module adheres closely to the IEC
61850 7-2 standard for describing AC measurements.
10.2. Power Levels
There are twelve Power Monitor Levels. They are subdivided into
six operational states, and six non-operational states. The
lowest non-operational state is 1 and the highest is six. Each
non-operational state corresponds to an ACPI level [ACPI].
<Claise, et. Al> Expires April 20, 2011 [Page 21]
Internet-Draft <Power Management Archictecure> Octobre 2010
11. Implementation Scenarios
The scope of power and energy monitoring consists of devices
that consume power within and that are connected to a
communications network. These devices include:
- Network devices and sub-components: Devices such as routers
and switches and their sub-components.
- Network attached endpoints: Devices that use the
communications network, such as endpoints, PCs, and facility
gateways that proxy energy monitor and control for commercial
buildings or home automation.
- Network attached meters or supplies: Devices that can monitor
the electrical supply, such as smart meters or Universal
Power Supplies (UPS) that meter and provide availability.
-
This section provides illustrative examples that model different
scenarios for implementation of the Power Monitor, including
Power Monitor Parent and Power Monitor Child relationships.
Each of the scenarios below is explained in more detail in the
Power Monitor MIB document [POWER-MON-MIB], with a mapping to
the MIB Objects.
Scenario 1: Switch with PoE endpoints
Consider a PoE IP phone connected to a switch. The IP phone
draws power from the PoE switch.
Scenario 2: Switch with PoE endpoints with further connected
device(s)
Consider the same example as in Scenario 1, but with a PC daisy-
chained from the IP phone for LAN connectivity. The phone draws
power from the PoE port of the switch, while the PC draws power
from the wall outlet.
Scenario 3: A switch with Wireless Access Points
Consider a WAP (Wireless Access Point) connected to the PoE port
of a switch. There are several PCs connected to the Wireless
Access Point over Wireless protocols. All PCs draw power from
the wall outlets.
<Claise, et. Al> Expires April 20, 2011 [Page 22]
Internet-Draft <Power Management Archictecure> Octobre 2010
The switch port is the Power Monitor Parent for the Wireless
Access Point (WAP) and all the PCs. But there is a distinction
among the Power Monitor Children, as the WAP draws power from
the PoE port of the switch and the PCs draw power from the wall
outlet.
Scenario 4: Network connected facilities gateway
At the top of the network hierarchy of a building network is a
gateway device that can perform protocol conversion between many
facility management devices, such as BACNET, MODBUS, DALI, LON,
etc. There are power meters associated with power-consuming
entities (Heating Ventilation & Air Conditioning - HVAC,
lighting, electrical, fire control, elevators, etc). The
proposed MIB can be implemented on the gateway device. The
gateway can be considered as the Power Monitor Parent, while the
power meters associated with the energy consuming entities can
be considered as its Power Monitor Children.
Scenario 5: Data center network
A typical data center network consists of a hierarchy of
switches. At the bottom of the hierarchy there are servers
mounted on a rack, and these are connected to the top-of-the-
rack switches. The top switches are connected to aggregation
switches that are in turn connected to core switches. As an
example, Server 1 and Server 2 are connected to different switch
ports of the top switch.
The proposed MIB can be implemented on the switches. The switch
can be considered as the Power Monitor Parent. The servers can
be considered as the Power Monitor Children.
Scenario 6: Building gateway device
Similar scenario as the scenario 4.
Scenario 7: Power consumption of UPS
Data centers and commercial buildings can have Uninterruptible
Power Supplies (UPS) connected to the network. The Power Monitor
can be used to model a UPS as a Power Monitor Parent with the
connected devices as Power Monitor Children.
<Claise, et. Al> Expires April 20, 2011 [Page 23]
Internet-Draft <Power Management Archictecure> Octobre 2010
Scenario 8: Power consumption of battery-based devices
A PC is a typical example of a battery-based device.
12. Security Considerations
Regarding the data attributes specified here, some or all may be
considered sensitive or vulnerable in some network environments.
Reading or writing these attributes without proper protection
such as encryption or access authorization may have negative
effects on the network capabilities.
12.1. Security Considerations for SNMP
Readable objects in a MIB modules (i.e., objects with a MAX-
ACCESS other than not-accessible) may be considered sensitive or
vulnerable in some network environments. It is thus important
to control GET and/or NOTIFY access to these objects and
possibly to encrypt the values of these objects when sending
them over the network via SNMP.
The support for SET operations in a non-secure environment
without proper protection can have a negative effect on network
operations. For example:
. Unauthorized changes to the Power Domain or business
context of a Power Monitor may result in misreporting or
interruption of power.
. Unauthorized changes to a power level may disrupt the power
settings of the different Power Monitors, and therefore the
level of functionality of the respective Power Monitors.
. Unauthorized changes to the demand history may disrupt
proper accounting of energy usage.
With respect to data transport SNMP versions prior to SNMPv3 did
not include adequate security. Even if the network itself is
secure (for example, by using IPsec), there is still no secure
control over who on the secure network is allowed to access and
GET/SET (read/change/create/delete) the objects in these MIB
modules.
It is recommended that implementers consider the security
features as provided by the SNMPv3 framework (see [RFC3410],
section 8), including full support for the SNMPv3 cryptographic
mechanisms (for authentication and privacy).
<Claise, et. Al> Expires April 20, 2011 [Page 24]
Internet-Draft <Power Management Archictecure> Octobre 2010
Further, deployment of SNMP versions prior to SNMPv3 is not
recommended. Instead, it is recommended to deploy SNMPv3 and to
enable cryptographic security. It is then a customer/operator
responsibility to ensure that the SNMP entity giving access to
an instance of these MIB modules is properly configured to give
access to the objects only to those principals (users) that have
legitimate rights to GET or SET (change/create/delete) them.
12.2. Security Considerations for IPFIX
EDITOR'S NOTE: to be completed if IPFIX is discussed in this
document
13. IANA Considerations
This document has no actions for IANA.
14. Acknowledgments
The authors would like to Michael Brown for improving the text
dramatically.
15. References
Normative References
[RFC3410] Case, J., Mundy, R., Partain, D., and B. Stewart,
"Introduction and Applicability Statements for Internet
Standard Management Framework ", RFC 3410, December
2002.
[RFC5101] B. Claise, Ed., Specification of the IP Flow
Information Export (IPFIX) Protocol for the Exchange of
IP Traffic Flow Information, RFC 5101, January 2008.
[POWER-MON-REQ] Quittek, J., Winter, R., Dietz, T., Claise, B.,
and M. Chandramouli, "Requirements for Power
Monitoring", draft-quittek-power-monitoring-
requirements-01 (work in progress), July 2010.
[POWER-MON-MIB] Claise, B., Chandramouli, M., Parello, J., and
Schoening, B., "Power and Energy Monitoring MIB",
<Claise, et. Al> Expires April 20, 2011 [Page 25]
Internet-Draft <Power Management Archictecure> Octobre 2010
draft-claise-energy-monitoring-mib-06, (work in
progress), October 2010.
Informative References
[RFC2863] McCloghrie, K., Kastenholz, F., "The Interfaces Group
MIB", RFC 2863, June 2000.
[RFC3444] Pras, A., Schoenwaelder, J. "On the Differences
between Information Models and Data Models", RFC 3444,
January 2003.
[ACPI] "Advanced Configuration and Power Interface
Specification", http://www.acpi.info/spec30b.htm
[LLDP] IEEE Std 802.1AB, "Station and Media Control
Connectivity Discovery", 2005.
[LLDP-MED-MIB] ANSI/TIA-1057, "The LLDP Management Information
Base extension module for TIA-TR41.4 media endpoint
discovery information", July 2005.
[DASH] "Desktop and mobile Architecture for System Hardware",
http://www.dmtf.org/standards/mgmt/dash/
Authors' Addresses
Benoit Claise
Cisco Systems, Inc.
De Kleetlaan 6a b1
Diegem 1813
BE
Phone: +32 2 704 5622
Email: bclaise@cisco.com
John Parello
Cisco Systems, Inc.
3550 Cisco Way
San Jose, California 95134
US
<Claise, et. Al> Expires April 20, 2011 [Page 26]
Internet-Draft <Power Management Archictecure> Octobre 2010
Phone: +1 408 525 2339
Email: jparello@cisco.com
Brad Schoening
Cisco Systems, Inc.
3550 Cisco Way
San Jose, California 95134
US
Phone: +1 408 525 2339
Email: braschoe@cisco.com
<Claise, et. Al> Expires April 20, 2011 [Page 27]