Internet DRAFT - draft-bagnulo-ippm-new-registry
draft-bagnulo-ippm-new-registry
Network Working Group M. Bagnulo
Internet-Draft UC3M
Intended status: Standards Track T. Burbridge
Expires: January 15, 2014 BT
S. Crawford
SamKnows
P. Eardley
BT
A. Morton
AT&T Labs
July 14, 2013
A registry for commonly used metrics
draft-bagnulo-ippm-new-registry-01
Abstract
This document creates a registry for commonly used metrics, defines
the rules for assignments in the new registry and performs initial
allocations. This document proposes one particular registry
structure with a single registry with multiple sub-registries. A
companion document draft-bagnulo-ippm-new-registry-independent
explores an alternative structure with independent registries for
each of the fields involved. .
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 http://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 January 15, 2014.
Bagnulo, et al. Expires January 15, 2014 [Page 1]
Internet-Draft Metrics Registry July 2013
Copyright Notice
Copyright (c) 2013 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.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3
2. The commonly used metrics registry . . . . . . . . . . . . . 4
2.1. The metrics registry . . . . . . . . . . . . . . . . . . 5
2.2. The Scheduling registry . . . . . . . . . . . . . . . . . 5
2.3. The Environment registry . . . . . . . . . . . . . . . . 6
2.4. The Output type registry . . . . . . . . . . . . . . . . 6
3. Initial assignment for the Scheduling registry . . . . . . . 6
3.1. Common parameter definitions . . . . . . . . . . . . . . 6
3.2. Poisson scheduling . . . . . . . . . . . . . . . . . . . 7
3.3. Periodic scheduling . . . . . . . . . . . . . . . . . . . 7
3.4. Singleton scheduling . . . . . . . . . . . . . . . . . . 8
4. Initial assignments for the Output Type registry . . . . . . 8
4.1. Raw . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.2. Xth percentile interval . . . . . . . . . . . . . . . . . 9
4.3. Xth percentile mean . . . . . . . . . . . . . . . . . . . 9
5. Initial assignments for the Environment registry . . . . . . 9
5.1. Undefined . . . . . . . . . . . . . . . . . . . . . . . . 9
5.2. No cross traffic . . . . . . . . . . . . . . . . . . . . 10
6. Initial assignments for the Metric registry . . . . . . . . . 11
6.1. Comment . . . . . . . . . . . . . . . . . . . . . . . . . 11
6.2. UDP related metrics . . . . . . . . . . . . . . . . . . . 11
6.2.1. No cross traffic, raw, Poisson, UDP latency metric . 12
6.2.2. No cross traffic, 99th percentile mean, Poisson, UDP
latency metric . . . . . . . . . . . . . . . . . . . 13
6.2.3. No cross traffic, 99th percentile interval, Poisson,
UDP latency metric . . . . . . . . . . . . . . . . . 14
6.2.4. No cross traffic, Poisson UDP packet-loss ratio
metric . . . . . . . . . . . . . . . . . . . . . . . 15
6.3. ICMP related metrics . . . . . . . . . . . . . . . . . . 16
6.3.1. No cross traffic, Periodic, ICMP packet-loss ratio
metric . . . . . . . . . . . . . . . . . . . . . . . 16
Bagnulo, et al. Expires January 15, 2014 [Page 2]
Internet-Draft Metrics Registry July 2013
6.4. DNS related metrics . . . . . . . . . . . . . . . . . . . 17
6.4.1. DNS latency metric . . . . . . . . . . . . . . . . . 18
6.5. VoIP related metrics . . . . . . . . . . . . . . . . . . 19
6.5.1. No cross traffic, raw, Periodic, UDP latency metric . 20
6.5.2. No cross traffic, raw, Periodic, UDP loss metric . . 21
6.5.3. No cross traffic, raw, Periodic, UDP, PDV metric . . 22
6.5.4. No cross traffic, Periodic, UDP PDV:99.9 metric . . . 24
6.5.5. No cross traffic, Periodic UDP packet-loss ratio
metric . . . . . . . . . . . . . . . . . . . . . . . 25
7. Security considerations . . . . . . . . . . . . . . . . . . . 26
8. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 26
9. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 26
10. References . . . . . . . . . . . . . . . . . . . . . . . . . 26
10.1. Normative References . . . . . . . . . . . . . . . . . . 26
10.2. Informative References . . . . . . . . . . . . . . . . . 27
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 27
1. Introduction
This document creates a registry for commonly used metrics. In order
to do that, it creates a number of namespaces whose values will be
recorded by the registry and will uniquely and precisely identify
metrics.
The motivation for having such registry is to allow a controller to
request a measurement agent to execute a measurement using a specific
metric. Such request can be performed using any control protocol
that refers to the value assigned to the specific metric in the
registry. Similarly, the measurement agent can report the results of
the measurement and by referring to the metric value it can
unequivocally identify the metric that the results correspond to.
There was a previous attempt to define a metric registry RFC 4148
[RFC4148]. However, it was obsoleted by RFC 6248 [RFC6248] because
it was "found to be insufficiently detailed to uniquely identify IPPM
metrics... [there was too much] variability possible when
characterizing a metric exactly" which led to the RFC4148 registry
having "very few users, if any".
Our approach learns from this, by tightly defining each entry in the
registry with only a few parameters open for each. The idea is that
the entries in the registry represent different measurement tests,
whilst the parameters set things like source and destination
addresses that don't change the fundamental nature of the test. The
downside of this approach is that it could result in an explosion in
the number of entries in the registry. We believe that less is more
in this context - it is better to have a reduced set of useful
metrics rather than a large set of metrics with questionable
Bagnulo, et al. Expires January 15, 2014 [Page 3]
Internet-Draft Metrics Registry July 2013
usefulness. Therefore this document defines that the registry only
includes commonly used metrics that are well defined; hence we
require both specification required AND expert review policies for
the assignment of values in the registry.
There are a couple of side benefits of having such registry. First
the registry could serve as an inventory of useful and used metrics,
that are normally supported by different implementations of
measurement agents. Second, the results of the metrics would be
comparable even if they are performed by different implementations
and in different networks, as the metric is properly defined.
The registry forms part of a Measurement Plan {do you prefer the term
'Characterization Plan', 'control framework' or 'test schedule'?}. It
describes various factors that need to be set by the party
controlling the measurements, for example: specific values for the
parameters associated with the selected registry entry (for instance,
source and destination addresses); and how often the measurement is
made. The Measurement Plan might look something like: "Dear
measurement agent: Please start test DNS(example.com) and
RTT(server.com,150) every day at 2000 GMT. Run the DNS test 5 times
and the RTT test 50 times. Do that when the network is idle.
Generate both raw results and 99th percentile mean. Send measurement
results to collector.com in IPFIX format". The Measurement Plan
depends on the requirements of the controlling party. For instance
the broadband consumer might want a one-off measurement made
immediately to one specific server; a regulator might want the same
measurement made once a day until further notice to the 'top 10'
servers; whilst an operator might want a varying series of tests
(some of which will be beyond those defined in the registry) as
determined from time to time by their operational support system.
While the registries defined in this document help to define the
Measurement Plan its full specification falls outside the scope of
this document.
2. The commonly used metrics registry
In this section we define the registry for commonly used metrics. It
is composed by the following sub-registries:
o Scheduling registry
o Environment registry
o Output-type registry
o Metric registry
Bagnulo, et al. Expires January 15, 2014 [Page 4]
Internet-Draft Metrics Registry July 2013
The rationale for the registry structure is to allow flexibility but
yet precise definition of metrics. The metric registry is the
fundamental registry and the other are auxiliary registries that
define values for different fields in the metric registry.
2.1. The metrics registry
Each entry for the metrics registry contain the following
information:
o Identifier: A text string that uniquely identifies the metric
o Name: The name of the metric
o Environment: A value from the Environment registry
o Output-type: A value from the Output-type registry
o Scheduling: A value form the Scheduling registry
o Reference: The specification where the metric is defined
The policy for the assignments in the metric registry is both
specification required AND expert review. This means that in order
to create an entry for the metric value a specification defining the
metric is required and when that happens, the request for allocation
will be reviewed by an expert.
The specification must define the input parameters for the metric as
well as the output of the metric. The metric must be well defined,
in the sense that two independent implementations must produce
uniform and comparable results.
The expert review must make sure that the proposed metric is
operationally useful. This means that the metric has proven to be
useful in operational/real scenarios.
2.2. The Scheduling registry
Each entry for the scheduling registry contain the following
information:
o Value: The name of the scheduling
o Reference: the specification where the scheduling is defined
The scheduling defines the scheduling strategy for the metric.
Simplest is Singleton scheduling, where an atomic measurement is
Bagnulo, et al. Expires January 15, 2014 [Page 5]
Internet-Draft Metrics Registry July 2013
made. Other strategies make a series of atomic measurements in a
"sample" or "stream", with the schedule defining the timing between
each distinct measurement. Each atomic measurement could consist of
sending a single packet (such as a DNS request) or sending several
packets (for example a webpage). A scheduling strategy requires
input parameter(s). Assignment in this registry follows the
specification required policy.
2.3. The Environment registry
Each entry for the environment registry contain the following
information:
o Value: The name of the environment
o Reference: the specification where the environment is defined
The environment defines the conditions where the metric is expected
to be used. It does not define the metric itself, but the context
where the metric is executed. Assignment in this registry follows
the specification required policy.
2.4. The Output type registry
Each entry for the output type registry contain the following
information:
o Value: The name of the output type
o Reference: the specification where the output type is defined
The output type define the type of output that the metric produces.
It can be the raw results or it can be some form of statistic.
Assignment in this registry follows the specification required
policy. The specification of the output type must define the format
of the output. Note that if two types of statistic are required from
the same test (for example, both "Xth percentile mean" and "Raw")
then a new output type must be defined ("Xth percentile mean AND
Raw").
3. Initial assignment for the Scheduling registry
3.1. Common parameter definitions
Although each IPPM RFC defines individual parameters and uses them
consistently, the parameter names are not completely consistent
across the RFC set. For example, the variable "dT" is used in
several different ways. This memo uses one set of parameter names,
Bagnulo, et al. Expires January 15, 2014 [Page 6]
Internet-Draft Metrics Registry July 2013
and the reader is cautioned to map the names according to their
definitions.
We define some parameters that are used by several types of
scheduling:
o T0: time to begin a test
o Tf: time to end a test
T0 and Tf are both in seconds and use the date (yyyy-mm-dd) and NTP
64 bit timestamp. T0 includes any control handshaking before the
test stream or singleton. Tf is the time the last test data is sent.
As a result, we have:
o Time when test devices may close the test socket: Tf + Waiting
Time (the time to wait before declaring a packet lost is fixed for
each metric)
o Total duration of the test: Tf - T0 + Waiting Time
3.2. Poisson scheduling
The values for this entry are as follows:
o Value: Poisson
o Reference: draft-bagnulo-ippm-new-registry
The Poisson scheduling is defined in section 11.1.1 of RFC 2330
[RFC2330] and needs input parameters:
o T0 and Tf: defined above
o lambda: the parameter defining the Poisson distribution. Lambda
is the mean number of distinct measurements per second in the
sample.
3.3. Periodic scheduling
The values for this entry are as follows:
o Value: Periodic
o Reference: draft-bagnulo-ippm-new-registry
Bagnulo, et al. Expires January 15, 2014 [Page 7]
Internet-Draft Metrics Registry July 2013
The Periodic sampling is defined in RFC 3432 [RFC3432]. The
additional input parameters for the metric required by Periodic
scheduling are:
o T0 and Tf: defined above
* Note that with Periodic sampling, T0 MUST NOT be strictly
periodic with other tests of the same type. RFC 3432 [RFC3432]
requires randomized start times and describes one way to
accomplish this. Also, the duration of the test MUST be
limited.
o incT: the time in seconds between one distinct event and the next,
where events typically result in repeating singleton measurements
of various types (illustrated below).
* for a periodic stream this is the time between packets in the
sample, first bit to first bit
* for measurements on a process this is the time between the
first packets of the process, for example first bit to first
bit of the SYN in a TCP 3-way handshake
3.4. Singleton scheduling
The values for this entry are as follows:
o Value: singleton
o Reference: draft-bagnulo-ippm-new-registry
The singleton scheduling covers the case when an atomic metric is
performed as per RFC 2330 [RFC2330]. The additional input parameter
for the metric required by Singleton scheduling is:
o T0: defined above
4. Initial assignments for the Output Type registry
4.1. Raw
The values for this entry are as follows:
o Value: Raw
o Reference: draft-bagnulo-ippm-new-registry
Bagnulo, et al. Expires January 15, 2014 [Page 8]
Internet-Draft Metrics Registry July 2013
The results of the metric are delivered in the exact way they are
produced by the measurements without any further processing or
filtering.
4.2. Xth percentile interval
The values for this entry are as follows:
o Value: Xth percentile interval
o Reference: draft-bagnulo-ippm-new-registry
The additional input parameter for the metric is:
o X: the percentile (e.g, if the X input parameter is 99, then the
output will be the 99th percentile interval.)
The output when using this Output type will be a a couple of values,
expressed in the same units as the raw output, that is the Xth
percentile interval, as defined in section 1.3 of RFC 2330 [RFC2330].
4.3. Xth percentile mean
The values for this entry are as follows:
o Value: Xth percentile mean
o Reference: draft-bagnulo-ippm-new-registry
The additional input parameter for the metric is
o X: the percentile (e.g, if the X input parameter is 99, then the
output will be the 99th percentile mean.)
The output when using this Output type will be a single value,
expressed in the same units as the raw output, that is the mean of
the sample only considering the values contained in the Xth
percentile interval, as defined in RFC 2330 [RFC2330].
5. Initial assignments for the Environment registry
5.1. Undefined
The values for this entry are as follows:
o Value: Undefined
o Reference: draft-bagnulo-ippm-new-registry
Bagnulo, et al. Expires January 15, 2014 [Page 9]
Internet-Draft Metrics Registry July 2013
The undefined environment is the case where no additional environment
settings are defined to perform the metric.
5.2. No cross traffic
The values for this entry are as follows:
o Value: No cross-traffic
o Reference: draft-bagnulo-ippm-new-registry
It is often important that there is no other traffic than the one
generated by the measurement itself while doing the measurement. The
reasons for this are two-folded, first, it is sometimes important
that the traffic created by the measurement doesn't impact the
experience of the users of the measured resource. Second it is
sometimes important that no other traffic interferes with the
measurement. This can be ensured by checking that the level of user
traffic is either zero or low enough to be confident that it won't
impact or be impacted by the measurement.
The "No cross traffic" condition is satisfied when, during the 5
seconds preceding measurement of the metric:
o the level of traffic flowing through the interface that will be
used to send measurement packets in either direction is less than
a threshold value of 1% of the line rate of the aforementioned
interface.
The "cross traffic" measurement is made at the interface, associated
with the measurement agent, that user traffic flows across. For
example, if the probe is attached to the home gateway, then the
interface is the service demarcation point where the subscriber
connects their private equipment or network to the subscribed
service.
Note that the No-cross traffic condition is defined only for the link
directly attached to the measurement agent initiating the
measurement. There is nothing mentioned about cross traffic on other
parts of the path used by measurement packets. In the case the
bottleneck of the path is other link than the one directly attached
to the device running the measurement agent, it may affect and be
affected by the measurement even if the No cross traffic as defined
here holds.
DISCUSSION
Bagnulo, et al. Expires January 15, 2014 [Page 10]
Internet-Draft Metrics Registry July 2013
o clarify whether traffic for each direction is less than threshold,
or the sum
o current SamKnows probes measure cross-traffic before the
measurement of the metric. Another approach would be to measure
cross-traffic during the time the metric is measured. Or a hybrid
approach. These would either be separate environment entries, or
parameterise the existing one.
o current SamKnows probes define a fixed threshold. it could be a
parameter
o could ignore broadcast traffic (think SamKnows includes)
o It would be possible to define this a bit more precisely as
follows:
The "No cross-traffic" condition is defined for active
measurements. The measurement agent runs in a device that has
one or more interfaces. In active measurements, the
measurement agent sends one or more packets. Lets call if0 the
interface through with the packets resulting from the
measurement are sent through. The no cross traffic condition
is fulfilled when during the 5 seconds prior sending each of
the packets of the measurement:
+ The traffic incoming through if0 that does not belong to the
measurement is lower than 1% of the line rate of if0
+ The traffic coming through the rest of the interfaces
towards if0 is less than 1% of the line rate of if0.
6. Initial assignments for the Metric registry
6.1. Comment
Need to work through that we only define T0 and Tf (and not T, dT).
6.2. UDP related metrics
RFC 2681 [RFC2681] defines a Round-trip delay metric and RFC 6673
[RFC6673] defines a Round-trip packet loss metric. We build on these
two metrics by specifying several of the open parameters to precisely
define several metrics for measuring UDP latency and packet loss.
All the UDP related metrics defined in this section use the
following:
P-Type:
Bagnulo, et al. Expires January 15, 2014 [Page 11]
Internet-Draft Metrics Registry July 2013
o IPv4 header values:
* DSCP: set to 0
* TTL set to 255
* Protocol: Set to 17 (UDP)
o UDP header values:
* Checksum: the checksum must be calculated
o Payload
* Sequence number: 8-byte integer
* Timestamp: 8 byte integer. Expressed as 64-bit NTP timestamp
as per section 6 of RFC 5905 [RFC5905]
* No padding
Timeout: 3 seconds
6.2.1. No cross traffic, raw, Poisson, UDP latency metric
We define the No cross traffic, raw, Poisson, UDP latency metric as
follows:
o Identifier: TBD1
o Name: No cross-traffic, raw, Poisson, UDP latency
o Environment: No cross-traffic, access measurement
o Output type: raw
o Scheduling: Poisson
o Reference: draft-bagnulo-ippm-new-registry
The methodology for this metric is defined as Type-P-Round-trip-
Delay-Poisson-Stream in RFC 2681 [RFC2681] using the P-Type and
Timeout defined above.
The input parameters for this metric are:
o Source IP Address
Bagnulo, et al. Expires January 15, 2014 [Page 12]
Internet-Draft Metrics Registry July 2013
o Destination IP Address
o Source UDP port
o Destination UDP port
o Initial time T
o Duration dT
o Rate lambda
The output of this metric is a list of elements. Each element
corresponds to one packet sent. Each element contains the timestamp
of the sent packet and the time when the echo was received.
6.2.2. No cross traffic, 99th percentile mean, Poisson, UDP latency
metric
The methodology for this metric is defined as Type-P-Round-trip-
Delay-Poisson-Stream in RFC 2681 [RFC2681] using the P-Type and
Timeout defined above. However, the output of the metric is not the
raw output, but the 99th percentile mean statistic.
The input parameters for this metric are:
o Identifier: TBD2
o Name: No cross-traffic, 99th percentile mean, Poisson, UDP latency
o Environment: No cross-traffic, access measurement
o Output type: Xth percentile mean
o Scheduling: Poisson
o Reference: draft-bagnulo-ippm-new-registry
The methodology for this metric is defined in RFC 2681 [RFC2681]
using the P-Type and Timeout defined above.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Source UDP port
Bagnulo, et al. Expires January 15, 2014 [Page 13]
Internet-Draft Metrics Registry July 2013
o Destination UDP port
o Initial time T
o Duration dT
o Rate lambda
o Xth percentile: 99
The output of this metric is a single value that corresponds to the
99th percentile mean of the results.
6.2.3. No cross traffic, 99th percentile interval, Poisson, UDP latency
metric
The methodology for this metric is defined as Type-P-Round-trip-
Delay-Poisson-Stream in RFC 2681 [RFC2681] using the P-Type and
Timeout defined above. However, the output of the metric is not the
raw output, but the 99th percentile interval statistic.
The input parameters for this metric are:
o Identifier: TBD3
o Name: No cross-traffic, 99th percentile interval, Poisson, UDP
latency
o Environment: No cross-traffic, access measurement
o Output type: Xth percentile interval
o Scheduling: Poisson
o Reference: draft-bagnulo-ippm-new-registry
The methodology for this metric is defined in RFC 2681 [RFC2681]
using the P-Type and Timeout defined above.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Source UDP port
o Destination UDP port
Bagnulo, et al. Expires January 15, 2014 [Page 14]
Internet-Draft Metrics Registry July 2013
o Initial time T
o Duration dT
o Rate lambda
o Xth percentile: 99
The output of this metric is a single value that corresponds to the
99th percentile interval of the results.
6.2.4. No cross traffic, Poisson UDP packet-loss ratio metric
We define the No cross traffic, Poisson, UDP packet-loss ratio metric
as follows:
o Identifier: TBD4
o Name: No cross-traffic, Poisson, UDP packet-loss ratio
o Environment: No cross-traffic, access measurement
o Output type: Xth percentile mean
o Scheduling: Poisson
o Reference: draft-bagnulo-ippm-new-registry
This metric is defined as Type-P-Round-trip-Loss-Poisson-Ratio in RFC
6673 [RFC6673] using the P-Type and Timeout defined above.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Source UDP port
o Destination UDP port
o Initial time T
o Duration dT
o Rate lambda
o X percentile: 100
Bagnulo, et al. Expires January 15, 2014 [Page 15]
Internet-Draft Metrics Registry July 2013
The output of this metric is a single value that corresponds to the
ratio of loss packets divided by the total number of packets sent.
6.3. ICMP related metrics
RFC 6673 [RFC6673] defines a Round-trip packet loss metric. We build
on that metrics by specifying several of the open parameters to
precisely define a metric for measuring ICMP packet loss. The ICMP
related metric defined in this document use the following:
P-Type:
o IPv4 header values:
* DSCP: set to 0
* TTL set to 255
* Protocol: Set to 1 (ICMP)
o ICMP header values:
* Type: 8 (Echo request)
* Code: 0
Observation: reply packets will contain an ICMP type of 0 Echo reply.
Timeout: 3 seconds
6.3.1. No cross traffic, Periodic, ICMP packet-loss ratio metric
We define the No cross traffic, Periodic, ICMP packet-loss ratio
metric as follows:
o Identifier: TBD6
o Name: No cross-traffic, Periodic, ICMP packet-loss ratio
o Environment: No cross-traffic, access measurement
o Output type: Xth percentile mean
o Scheduling: Periodic
o Reference: draft-bagnulo-ippm-new-registry
Bagnulo, et al. Expires January 15, 2014 [Page 16]
Internet-Draft Metrics Registry July 2013
This metric is defined as Type-P-Round-trip-Loss-Periodic-Ratio in
RFC 6673 [RFC6673] using the P-Type and Timeout defined above.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Initial time T
o End Time Tf
o dt (the interval allowed for sample start times)
o Inter-packet time: incT
o X percentile: 100
The output of this metric is a single value that corresponds to the
ratio of loss packets divided by the total number of packets sent.
6.4. DNS related metrics
RFC 2681 [RFC2681] defines a Round-trip delay metric. We build on
that metric by specifying several of the open parameters to precisely
define a metric for measuring DNS latency. The metric uses the
following parameters:
P-Type:
o IPv4 header values:
* DSCP: set to 0
* TTL set to 255
* Protocol: Set to 17 (UDP)
o UDP header values:
* Source port: 53
* Destination port: 53
* Checksum: the checksum must be calculated
Bagnulo, et al. Expires January 15, 2014 [Page 17]
Internet-Draft Metrics Registry July 2013
o Payload: The payload contains a DNS message as defined in RFC 1035
[RFC1035] with the following values:
* The DNS header section contains:
+ QR: set to 0 (Query)
+ OPCODE: set to 0 (standard query)
+ AA: not set
+ TC: not set
+ RD: set to one (recursion desired)
+ RA: not set
+ RCODE: not set
+ QDCOUNT: set to one (only one entry)
+ ANCOUNT: not set
+ NSCOUNT: not set
+ ARCOUNT: not set
* The Question section contains:
+ QNAME: the FQDN provided as input for the test
+ QTYPE: the query type provided as input for the test
+ QCLASS: set to IN
* The other sections do not contain any Resource Records.
Observation: reply packets will contain a DNS response and may
contain RRs.
Timeout: 3 seconds
6.4.1. DNS latency metric
We define the DNS latency metric as follows:
o Identifier: TBD7
Bagnulo, et al. Expires January 15, 2014 [Page 18]
Internet-Draft Metrics Registry July 2013
o Name: DNS latency
o Environment: Undefined
o Output type: raw
o Scheduling: Singleton
o Reference: draft-bagnulo-ippm-new-registry
The methodology for this metric is defined as Type-P-Round-trip-Delay
in RFC 2681 [RFC2681] using the P-Type and Timeout defined above.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address (the address of the DNS server to be
tested)
o QTYPE: A RR
o FQDN: a valid FQDN that will be queried for.
o Time T
The output of this metric is the timestamp when the packet was sent
and the delay that it took to receive a response. Please note that
any DNS response is valid, including no records in the answer.
(Should we be more explicit about what is the output when there is no
reply packet received?)
6.5. VoIP related metrics
[RFC2679] defines a one-way delay metric and [RFC2680] defines a one-
way packet loss metric. IPPM has derived a general packet delay
variation metric in [RFC3393], which we apply as recommended in
section 4.2 of [RFC5481]. We build on these specifications by
specifying several of the open parameters to precisely define several
metrics for measuring Voice over IP (VoIP) delay, delay variation,
and packet loss. All the VoIP related metrics defined in this
section use the following:
Type-P:
o IPv4 header values:
* DSCP: set to 0 (I think we move this to the sub-sections)
Bagnulo, et al. Expires January 15, 2014 [Page 19]
Internet-Draft Metrics Registry July 2013
* TTL set to 255
* Protocol: Set to 17 (UDP)
o UDP header values:
* Checksum: the checksum must be calculated
o Payload - suffcient octets to emulate a VoIP audio payload,
including the an RTP header if desired, the actual test protocol
will populate the payload with a measurement header containing
fields such as:
* Sequence number:
* Timestamp:
* Random bit pattern:
Waiting Time to declare a packet lost: 5 seconds
Periodic Stream Description:
o Nominal inter-packet interval incT=20ms (first bit to first bit)
6.5.1. No cross traffic, raw, Periodic, UDP latency metric
We define the No cross traffic, raw, Periodic, UDP latency metric as
follows:
o Identifier: TBD641
o Name: No cross-traffic, raw, Periodic, UDP latency
o Environment: No cross-traffic, access measurement
o Output type: raw
o Scheduling: Periodic
o Reference: draft-bagnulo-ippm-new-registry
The methodology for this metric is defined as Type-P-One-way-Delay-
Periodic-Stream in section 4 of [RFC3432], including parameters from
section 3 of [RFC3432] and using the Type-P and Waiting Time defined
above in section 6.4.
The input parameters for this metric are:
Bagnulo, et al. Expires January 15, 2014 [Page 20]
Internet-Draft Metrics Registry July 2013
o Source IP Address
o Destination IP Address
o Source UDP port
o Destination UDP port
o Beginning of testing interval, T
o Initial time T0 (including a random offset from the beginning of
T)
o Ending time Tf
Variable aspects of Type-P are:
o DSCP value
o UDP Payload length
The output of this metric is a list of elements. Each element
corresponds to one packet sent. Each element contains the timestamp
of the sent packet and the time when the packet was received at the
destination (from which the one-way delay can be calculated). The
methodology's sequence number MAY be included. For packets which do
not arrive prior to the Waiting Time, the received timestamp for that
packet SHOULD be indicated as "not available", or post-processing may
be applied to enforce the constant Waiting Time to exclude long-
delayed packets and lost packets from further analysis.
6.5.2. No cross traffic, raw, Periodic, UDP loss metric
We define the No cross traffic, raw, Periodic, UDP loss metric as
follows:
o Identifier: TBD642
o Name: No cross-traffic, raw, Periodic, UDP latency
o Environment: No cross-traffic, access measurement
o Output type: raw
o Scheduling: Periodic
o Reference: draft-bagnulo-ippm-new-registry
Bagnulo, et al. Expires January 15, 2014 [Page 21]
Internet-Draft Metrics Registry July 2013
The methodology for this metric is identical to Type-P-One-way-Delay-
Periodic-Stream in section 4 of [RFC3432], including parameters from
section 3 of [RFC3432] and using the Type-P and Waiting Time defined
above in section 6.4.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Source UDP port
o Destination UDP port
o Beginning of testing interval, T
o Initial time T0 (including a random offset from the beginning of
T)
o Ending time Tf
Variable aspects of Type-P are:
o DSCP value
o UDP Payload length
The output of this metric is a list of elements. Each element
corresponds to one packet sent. Each element contains the timestamp
of the sent packet and the time when the packet was received at the
destination (from which the one-way delay can be calculated). The
methodology's sequence number MAY be included. For packets which do
not arrive prior to the Waiting Time, the received timestamp for that
packet SHOULD be indicated as "not available", or post-processing may
be applied to enforce the constant Waiting Time to exclude long-
delayed packets and lost packets from further analysis.
Note that the same raw output format MAY serve both loss and delay
metrics.
6.5.3. No cross traffic, raw, Periodic, UDP, PDV metric
We define the No cross traffic, Periodic, UDP Packet Delay Variation
metric as follows:
o Identifier: TBD643
Bagnulo, et al. Expires January 15, 2014 [Page 22]
Internet-Draft Metrics Registry July 2013
o Name: No cross-traffic, Periodic, UDP PDV
o Environment: No cross-traffic, access measurement
o Output type: raw
o Scheduling: Periodic
o Reference: draft-bagnulo-ippm-new-registry
The methodology for the delay singletons from which this metric is
derived take the first steps defined as Type-P-One-way-Delay-
Periodic-Stream in section 4 of [RFC3432], including parameters from
section 3 of [RFC3432] and using the Type-P and Waiting Time defined
above in section 6.4. This collects the one-way delay singletons.
The next step in the methodology follows from sections 2 and 3 of
[RFC3393] (which describes how to use a selection function to
determine pairs of packets to derive PDV) and section 4.2 of
[RFC5481], where the packet with the minimum delay is specified as a
fixed member of the pair.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Source UDP port
o Destination UDP port
o Beginning of testing interval, T
o Initial time T0 (including a random offset from the beginning of
T)
o Ending time Tf
Variable aspects of Type-P are:
o DSCP value
o UDP Payload length
The output of this metric is a list of triples (3 elements). Each
element corresponds to one packet in the sample. Each element
contains the one way delay of the first packet in the pair, the one
Bagnulo, et al. Expires January 15, 2014 [Page 23]
Internet-Draft Metrics Registry July 2013
way delay of the second packet in the pair (having minimum delay),
and the variation in transmission time calculated for packet with
sequence number i as PDV(i) = D(i)-D(min). The methodology's
sequence number MAY be included. For packets which do not arrive
prior to the Waiting Time, the delay for that packet and its PDV
SHOULD be indicated as "not available" (following section 4.1 of
[RFC3393]).
6.5.4. No cross traffic, Periodic, UDP PDV:99.9 metric
We define the No cross traffic, Periodic, UDP Packet Delay Variation
(99.9 percentile) metric as follows:
o Identifier: TBD644
o Name: No cross-traffic, Periodic, UDP PDV:99.9
o Environment: No cross-traffic, access measurement
o Output type: 99.9 percentile
o Scheduling: Periodic
o Reference: draft-bagnulo-ippm-new-registry
The methodology for the delay singletons from which this metric is
derived take the first steps defined as Type-P-One-way-Delay-
Periodic-Stream in section 4 of [RFC3432], including parameters from
section 3 of [RFC3432] and using the Type-P and Waiting Time defined
above in section 6.4. This collects the one-way delay singletons.
The next step in the methodology follows from sections 2 and 3 of
[RFC3393] (which describes how to use a selection function to
determine pairs of packets to derive PDV) and section 4.2 of
[RFC5481], where the packet with the minimum delay is specified as a
fixed member of the pair.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Source UDP port
o Destination UDP port
o Beginning of testing interval, T
Bagnulo, et al. Expires January 15, 2014 [Page 24]
Internet-Draft Metrics Registry July 2013
o Initial time T0 (including a random offset from the beginning of
T)
o Ending time Tf
Variable aspects of Type-P are:
o DSCP value
o UDP Payload length
The output of this metric is a single value corresponding to the
99.9th percentile of PDV. For packets which do not arrive prior to
the Waiting Time, the delay for that packet and its PDV SHOULD be
indicated as "not available" (following section 4.1 of [RFC3393]).
If the 99.9th percentile of singletons corresponds to packet whose
delay and PDV are "not available", then the output of this metric is
"not available".
6.5.5. No cross traffic, Periodic UDP packet-loss ratio metric
We define the No cross traffic, Periodic, UDP packet-loss ratio
metric as follows:
o Identifier: TBD645
o Name: No cross-traffic, Periodic, UDP packet-loss ratio
o Environment: No cross-traffic, access measurement
o Output type: X percentile mean
o Scheduling: Periodic
o Reference: draft-bagnulo-ippm-new-registry
This metric is defined as Type-P-One-way-Loss-Average in Section 4.1
of[RFC2680] using the Type-P and Waiting Time defined in section 6.4
above.
The input parameters for this metric are:
o Source IP Address
o Destination IP Address
o Source UDP port
Bagnulo, et al. Expires January 15, 2014 [Page 25]
Internet-Draft Metrics Registry July 2013
o Destination UDP port
o Beginning of testing interval, T
o Initial time T0 (including a random offset from the beginning of
T)
o Ending time Tf
o X percentile mean: 100
Variable aspects of Type-P are:
o DSCP value
o UDP Payload length
The output of this metric is one value that corresponds to the ratio
of lost packets divided by the total number of packets sent. This
can be calculated from the singleton elements of section 6.4.2 above,
assigning the logical value "0" to packets with a valid one-way delay
and the value "1" to all packets whose one-way delay is recorded as
"not available". As section 4.1 of [RFC2680] indicates, the average
of all the logical values is the ratio of lost to total packets.
7. Security considerations
TBD
8. IANA Considerations
TBD
9. Acknowledgments
We would like to thank Henning Schulzrinne for many constructive
comments and input on early versions of this document.
10. References
10.1. Normative References
[RFC2330] Paxson, V., Almes, G., Mahdavi, J., and M. Mathis,
"Framework for IP Performance Metrics", RFC 2330, May
1998.
Bagnulo, et al. Expires January 15, 2014 [Page 26]
Internet-Draft Metrics Registry July 2013
[RFC3432] Raisanen, V., Grotefeld, G., and A. Morton, "Network
performance measurement with periodic streams", RFC 3432,
November 2002.
[RFC2681] Almes, G., Kalidindi, S., and M. Zekauskas, "A Round-trip
Delay Metric for IPPM", RFC 2681, September 1999.
[RFC6673] Morton, A., "Round-Trip Packet Loss Metrics", RFC 6673,
August 2012.
[RFC1035] Mockapetris, P., "Domain names - implementation and
specification", STD 13, RFC 1035, November 1987.
[RFC5905] Mills, D., Martin, J., Burbank, J., and W. Kasch, "Network
Time Protocol Version 4: Protocol and Algorithms
Specification", RFC 5905, June 2010.
[RFC2679] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way
Delay Metric for IPPM", RFC 2679, September 1999.
[RFC2680] Almes, G., Kalidindi, S., and M. Zekauskas, "A One-way
Packet Loss Metric for IPPM", RFC 2680, September 1999.
[RFC3393] Demichelis, C. and P. Chimento, "IP Packet Delay Variation
Metric for IP Performance Metrics (IPPM)", RFC 3393,
November 2002.
[RFC5481] Morton, A. and B. Claise, "Packet Delay Variation
Applicability Statement", RFC 5481, March 2009.
10.2. Informative References
[RFC4148] Stephan, E., "IP Performance Metrics (IPPM) Metrics
Registry", BCP 108, RFC 4148, August 2005.
[RFC6248] Morton, A., "RFC 4148 and the IP Performance Metrics
(IPPM) Registry of Metrics Are Obsolete", RFC 6248, April
2011.
Authors' Addresses
Bagnulo, et al. Expires January 15, 2014 [Page 27]
Internet-Draft Metrics Registry July 2013
Marcelo Bagnulo
Universidad Carlos III de Madrid
Av. Universidad 30
Leganes, Madrid 28911
SPAIN
Phone: 34 91 6249500
Email: marcelo@it.uc3m.es
URI: http://www.it.uc3m.es
Trevor Burbridge
British Telecom
Adastral Park, Martlesham Heath
Ipswich
ENGLAND
Email: trevor.burbridge@bt.com
Sam Crawford
SamKnows
Email: sam@samknows.com
Philip Eardley
British Telecom
Adastral Park, Martlesham Heath
Ipswich
ENGLAND
Email: philip.eardley@bt.com
Al Morton
AT&T Labs
200 Laurel Avenue South
Middletown, NJ
USA
Email: acmorton@att.com
Bagnulo, et al. Expires January 15, 2014 [Page 28]
