Technical Report No. 78, January 2003
Prepublished Version of T1.TR.78-2003
Technical Report on
Access Availability of Routers in
IP-Based Networks
Prepared by
T1A1.2
Working Group on
Network Survivability Performance
T1.TR.78-2003
Copyright ? 2003 by Alliance for Telecommunications Industry Solutions
All rights reserved.
Committee T1 is sponsored by the Alliance for Telecommunications Industry Solutions (ATIS) and accredited by the American National Standards Institute (ANSI). No part of this publication may be reproduced in any form, in an electronic retrieval system or otherwise, without the prior written permission of the publisher.
Technical Report
on
Access Availability of Routers in
IP-Based Networks
Abstract
This Technical Report (TR) introduces the concepts for use in assessing the availability of access to IP-based telecommunications networks. The calculation is based on the access availability of IP routers in the network. The TR presents alternate methods for weighting the availability calculation in terms of customers, ports, and bandwidth.
This Technical Report is intended as the first in a series of Technical Reports, on the reliability metrics for IP-based networks. The next report will include Backbone networks thereby permitting a complete network availability assessment.
T1.TR.78-2003
Prepared by
T1A1.2
Working Group on
Network Survivability Performance
Foreword
This draft Technical Report provides a practical way of assessing the availability of access to IP-based telecommunications networks. It is intended as a guide for network reliability practitioners. Future reports will address reliability metrics for IP backbone networks.
This draft technical report is intended for providers of IP-based telecommunications networks and services, and telecommunications equipment suppliers.
Suggestions for enhancement of this report are welcome. These should be sent to the Alliance for Telecommunications Industry Solutions, Suite 500, 1200 G Street N.W., Washington, D.C. 20005.
Working Group T1A1.2 on Network Survivability Performance, which developed this report, has the following officers and participants:
Chair:O. Avellaneda
Vice-Chair:S. Makris
Chief Editor: F. Kaudel
Editors:
J. Bennett Y. Kogan P. Tarapore
Active Participants:
O. Avellaneda
J. Bennett D. Clark W. Chiles C. Dvorak F. Kaudel Y. Kogan J. Lankford J. Lord A. McCain A. Nguyen R. Paterson J. Rupe A. Webster R. Wohlert
TECHNICAL REPORT NO. 78
Table of Contents
0EXECUTIVE SUMMARY (1)
1PURPOSE, SCOPE, APPLICATION, OUTLINE, AND DISCLAIMER (1)
1.1P URPOSE (1)
1.2S COPE (1)
1.3A PPLICATION (1)
1.4O UTLINE (1)
2INTRODUCTION (2)
3RELATED WORK (2)
3.1T1A1T ECHNICAL S UBCOMMITTEE (2)
3.2O THER T1T ECHNICAL S UBCOMMITTEES (2)
3.3I NTERNATIONAL S TANDARDS W ORK AT THE ITU-T (2)
3.4O THER F ORUMS AND C OMMITTEES (3)
4ACCESS NETWORK TOPOLOGY (3)
5IP ROUTER ACCESS AVAILABILITY CONCEPTS (5)
5.1IP R OUTER A CCESS T OTAL A VAILABILITY (5)
5.2IP R OUTER A CCESS A VAILABILITY (5)
5.3A VERAGE IP R OUTER A CCESS A VAILABILITY FOR A N ETWORK (6)
6CONCLUSIONS (7)
7FUTURE WORK AND REQUEST FOR USER/INDUSTRY FEEDBACK (7)
8BIBLIOGRAPHY (8)
9DEFINITIONS (8)
10ABBREVIATIONS AND ACRONYMS (9)
APPENDIX A: EXAMPLES OF IP ROUTER ACCESS AVAILABILITY CONCEPTS (10)
A.1E XAMPLE OF IP R OUTER A CCESS T OTAL A VAILABILITY (10)
A.2E XAMPLES OF IP R OUTER A CCESS A VAILABILITY (10)
A.3E XAMPLES OF A VERAGE IP R OUTER A CCESS A VAILABILITY FOR A N ETWORK (12)
APPENDIX B: TECHNIQUES FOR SCALING AVAILABILITY (14)
0 Executive Summary
This draft Technical Report (TR) provides a practical way of assessing the availability of access to IP-based telecommunications networks. It is intended as a guide for network reliability practitioners.
1 Purpose, Scope, Application, Outline, and Disclaimer
1.1Purpose
This TR was developed in the interest of relating useful availability measures for access to IP-based networks. In particular, there is no widely accepted way of relating IP component failures to service outages as experienced by users of IP services. This TR is intended to provide initial guidance in this area, with the goal of stimulating greater industry collaboration on IP reliability issues. This TR is complementary to other T1A1 work on IP performance of networks and services.
1.2 Scope
This TR provides a practical way of assessing the availability of access to IP networks. The use of these measures is demonstrated through examples.
This methodology is seen as being highly useful because it complements the traditional methods for reliability assessments. While highly practical, this method is one of several possible methods that could be used for assessing IP network reliability, and is not intended to preclude the use of other methodologies.
This TR addresses only IP-based access networks (and not IP-based backbone networks). Other than total loss of service from equipment outages, it does not include consideration of defects (such as delay, packet loss, and jitter) that may prevent satisfactory delivery of service. Work is ongoing at Working Group T1A1.2 in these areas. This TR will be useful to the industry with the understanding that these issues still need to be addressed.
1.3 Application
The guidance in this TR applies to IP access networks. Subsequent reports will address IP core networks and IP services and applications. This TR is intended as the first in a series.
1.4Outline
Section 0 provides an executive summary. Section 1 describes the Purpose, Scope, Application, and Outline. Section 2 provides an Introduction. Related Work is described in Section 3. A generic access network topology is described in Section 4. Section 5 defines availability concepts for access to IP networks. Section 6 presents the Conclusions. Section 7 describes Future Work. Section 8 presents a Bibliography. Definitions are presented in Section 9. Section 10 contains Acronyms and Abbreviations. Appendix A presents examples of calculating IP router access availability. Appendix B describes alternate equivalent forms for expressing availability.
2 Introduction
With the proliferation of technologies such as IP-based systems, there is an urgent need to be able to relate the overall quality requirements to the performance and reliability of the many underlying network and system elements. Yet to date there is no well-accepted method in the industry for relating failures in network elements to service-level outages, so this report is a start in this much-needed direction. Ultimately, all performance and reliability outages should be expressible in terms of their impacts on the users of a service.
3 Related Work
3.1T1A1 Technical Subcommittee
The T1A1 Technical Subcommittee comprises two Working Groups:
?T1A1.2 – Network Survivability Performance
?T1A1.3 – Performance of Networks and Services.
This TR is a product of the work underway in the T1A1.2 Working Group. This document supports the Technical Report No. 70 on “Reliability/Availability Framework for IP-based Networks and Services”[1]. Section 8 of the latter document deals with “Guides and Metrics” for assessing network reliability.
The T1A1.2 Working Group (WG) is responsible for the following Technical Reports (TR): ?TR No. 42 – “Enhanced Analysis of FCC-Reportable Service Outages”[2]
?TR No. 55 – “Reliability and Survivability Aspects of the Interactions between the Internet and the Public Telecommunications Network”[3]
?TR No. 68 – “Enhanced Network Survivability Performance”[4].
3.2 Other T1 Technical Subcommittees
Other T1 technical subcommittees involved in network reliability work include: ?T1E1 – Interfaces, Power, and Protection for Networks
?T1M1 – Internetwork Operations, Administration, Maintenance, and Provisioning
?T1S1 – Services, Architectures, and Signaling
?T1X1 – Digital Hierarchy and Synchronization.
3.3 International Standards Work at the ITU-T
Some of the ITU-T study groups involved in work related to network survivability performance are listed below:
?Study Group 2 – Operational Aspects of Service Provision, Networks, and Performance[5]
?Study Group 12 – End-to-End Transmission Performance of Networks and Terminals[6]
?Study Group 13 – Multi-Protocol and IP-based Networks and their Interworking
?Study Group 15 – Optical and Other Transport Networks.
3.4 Other Forums and Committees
Forums involved in network reliability include the following[1]:
?IETF – Internet Engineering Task Force
?NRIC – Network Reliability and Interoperability Council
?NRSC – Network Reliability Steering Committee
?OIF – Optical Internetworking Forum.
4 Access Network Topology
This section describes a fairly general IP Access Network. This network comprises all elements responsible for delivering transactions from the Customer Premise Equipment (CPE) at a customer location into the IP network backbone.
One common component to all these elements is the Customer Port. These ports are on the “drop side” of an Access Router, where facilities from a customer’s CPE terminate. A failure in any element in the Access Network may result in downtime for customer ports on the Access Routers. Such failures prevent delivery of customer transactions to the backbone. There are five (5) element types in a typical IP Access Network topology (Figure 1) whose failure can cause downtime for customer ports:
?Facilities and supporting elements such as cross-connects, which link routers at a customer’s CPE to ISP Access Router customer ports. Note that a customer may also use redundancy (double homing) by buying access to more than one Access Router of the same ISP or even different ISPs. However, this type of redundancy is not counted in the availability calculation because customers pay separately for each access port.
?Access Routers that form an edge on an ISP backbone network. Customer port downtime can be caused by a failure in a router component, such as a failed interface card, or from a total router failure.
?Facilities and supporting elements such as cross-connects, which link Access Routers to Backbone Routers. To increase the availability of the Access Network, an ISP usually provides redundancy by connecting each Access Router to two Backbone Routers at the same access node using two independent sets of uplinks1(Figure 1 depicts a typical access node with several Access Routers and two Backbone Routers). This permits customer traffic to enter the backbone in the following failure scenarios:
- A failed uplink
- A failed card supporting an uplink
- A failed Backbone Router at the access node.
1 An uplink is a facility connecting any router to a backbone router.
?Backbone Routers linked to Access Routers. As shown in Figure 1, if both Backbone Routers at an access node fail (a rare event), then all Access Routers at this node lose connection to the backbone.
?Facilities linking Backbone Routers at an access node, to backbone routers at other backbone nodes. Such facility failures decrease the available bandwidth from Access Routers to the backbone. Note that if all Backbone Router uplinks at an access node fail
(a rare event), then all Access Routers at this node lose connection to the backbone.
Figure 1: Access Network Elements
KEY
AR: Access Router
BR: C: Backbone Router Customer Port
As mentioned above, the common denominator for all failure types is the customer port on the Access Router. Any one of these failures may result in the “unavailability” of some of these Access Router customer ports. Customer port “unavailability” can be caused by a direct port failure or from the failure of other elements described above (Example: both Backbone Routers failed at the access node as shown in Figure 1).
From the perspective of network access, the Access Router (see Figure 1) is the key element of the Access Network. These routers receive IP packets from a customer location and route them into the selected network Backbone Router. Thus, any failure in the Access Network may result in downtime for some, or all, of the customer ports in the Access Router over the failure duration (usually measured in hours). For example, if a facility (or a facility supporting element such as a cross-connect) connecting the customer office location to the Access Router fails, then the corresponding Access Router port where the failed facility terminates, is considered to be unavailable over the failure duration. Section 5 defines availability concepts for access to these routers.
5
IP Router Access Availability Concepts 5.1
IP Router Access Total Availability
Given: ? A router is monitored for outages in a time period from time t 1 to time t 2,
? In that time period the router is in-service for a length of time T,
? N outages are reported in that time period, and
? The duration D is recorded for each outage, then
the availability of the router in time period [t 1, t 2] is Total access availability of router T D 1N 1i i
=?=.
The total availability of the router is a measure of the fraction of the in-service time that the router was not impacted by an outage.
5.2 IP Router Access Availability
The definition given above treats each outage as a total outage, one that denies access to all services for all customers. Router access availability may be defined with greater precision if each outage duration is weighted by the fraction of router service affected by the outage. Given:
? A router is monitored for outages in a time period from time t 1 to time t 2,
? In that time period the router is in-service for a length of time T,
? N outages are reported in that time period,
? The duration D is recorded for each outage, and
? The fraction f of the router access impacted by the outage is recorded for each outage,
then
the access availability of the router in time period [t 1, t 2] is Access availability of router T D f 1N 1i i
i
=?=.
The fraction f of a router impacted by an outage may be examined in many ways. Three of the most intuitive techniques for calculating the fraction are based on:
? Customers f =
router the by served customers of Number outage the by impacted customers of Number ? Ports f =
router the on service in ports access of Number outage the by impacted ports access of Number ? Bandwidth
Given a router with P access ports in-service having bandwidths b j , the total access bandwidth for the router may be defined as
B = =P 1j j b
.
The fraction of access to the router impacted by an outage is f = B b x P 1
j j
j =
where x j = 1 if access port j was impacted by the outage and 0 otherwise.
5.3
Average IP Router Access Availability for a Network
Given: ? A network of R routers
? Each router has been monitored for outages over the same time period [t 1, t 2]
? The access availability A of each router has been calculated over that period
? Each router has a weight w such that the weights of all routers sum to 1, then
the average access availability of the network routers in time period [t 1, t 2] is
Average access availability of network routers = =R 1k k k A w
.
As in the methods for determining partial impact of an outage, three of the most intuitive techniques for calculating router weights within a network are based on:
? Customers
If router r was in-service for a length of time T r and served C r customers in time period
[t 1, t 2], then
w r = =R 1k k
k r
r T C
T C . ? Ports
If router r had P r access ports and was in-service for a length of time T r in time period [t 1, t 2], then
w r = =R 1
k k k
r
r T P T P .
? Bandwidth
If router r had total access bandwidth B r and was in-service for a length of time T r in time period [t 1, t 2], then
w r = =R 1k k
k r
r T B
T B . 6 Conclusions
With the proliferation of IP-based systems, networks and services, there is an urgent need for methods to evaluate their availability in a production environment. To date, there is no well-accepted method in the industry for relating failures in network elements to service outages. The availability concepts introduced in this TR are an attempt to fill the gap in this much-needed area.
7 Future Work and Request for User/Industry Feedback
Ultimately, any performance and reliability degradation should be expressible in terms of the impact that such degradation has on the users of a service. With this goal in mind, T1A1 plans to continue work on concepts to assess the reliability of IP networks and services.
In view of this ongoing work, members of the industry and members of the user community are encouraged to provide information and feedback to T1A1 so that we can include your needs in formulating the direction and content of our work. In particular, information on user expectations regarding the performance and reliability of IP systems is especially solicited.
8 Bibliography
[1] “Reliability/Availability Framework for IP-based Networks and Services”, Technical
Report No. 70, T1A1.2 Working Group (Network Survivability Performance),
September 2001.
[2] “Enhanced Analysis of FCC-Reportable Service Outages”, Technical Report No. 42,
T1A1.2 Working Group (Network Survivability Performance), August 1995.
[3] “Reliability and Survivability Aspects of the Interactions between the Internet and the
Public Telecommunications Network”, Technical Report No. 55, T1A1.2 Working Group (Network Survivability Performance).
[4] “Enhanced Network Survivability Performance”, Technical Report No. 68, T1A1.2
Working Group (Network Survivability Performance), January 2001.
[5] “Customer Affecting Incidents and Blocking Defects Per Million”, ITU-T
Recommendation E.436, March 1998.
[6] “General Aspects of Quality of Service and Network Performance in Digital Networks”
ITU-T Recommendation I.350, March 1993.
[7] “Reliability and Quality Measurements for Telecommunications System (RQMS –
Wireline)”, Telcordia GR-929-CORE, Issue 7, December 2001.
[8] “Defects per Million as a Measure of IP Network Reliability Part 1: Access Networks”,
Y. Kogan, AT&T, presented as contribution T1A1.2/2001-028, Working Group T1A1.2 (Network Survivability Performance), Tucson, AZ, April 2001.
[9] “Communications Quality of Service: A Framework and Definitions”, ITU-T
Recommendation G.1000, November 2001.
9 Definitions
Access Networks The portion of a network over which telecommunications traffic
enters a service provider’s domain for routing and processing
towards the desired destination, and then proceeds into the
provider’s backbone.
Availability Customer Ports The probability that a system or element can perform its required functions.
Ports on the “drop side” of an Access Router linked to a customer’s office.
Uplink For any router (access or backbone), an uplink is a facility that
connects it to a backbone router.
10 Abbreviations and Acronyms
ATM Asynchronous Transfer Mode
DS-n Digital Signal – level n
IP Internet Protocol
ISP Internet Service Provider
Mbps Megabits per second (Million bits per second) OC-n Optical Carrier – level n
SLA Service Level Agreement
VoIP Voice over IP
Appendix A: Examples of IP Router Access Availability Concepts
The examples in this appendix illustrate the availability concepts described in Section 5. The intent is to provide a working familiarity with these concepts, and hence, these examples are relatively simple in form. More complex examples involving large numbers of access routers can be found in the listed references [8].
A.1 Example of IP Router Access Total Availability
A router is monitored for outages over a time period in which it is in-service for T = 1,000 hours. In that time period, N = 3 outages occur with durations of D 1 = 1 hour, D 2 = 0.4 hours, and D 3 = 0.6 hours. The total access availability of the router in this period is
Total access availability of router 998.1000
6.04.011T D
1N 1i i
=++?=?= =. A.2 Examples of IP Router Access Availability
Three examples are given for calculating IP router access availability. Each example uses a different basis (customer, port, bandwidth) for calculating the fraction of the router impacted by each outage. All three examples assume that the router is monitored for outages over a time period in which it is in-service for T = 1,000 hours.
Example of IP Router Access Availability on a Customer Basis
The router has 100 customers. In the time period, three outages occurred as indicated below: Outage
Duration D Customers Impacted Fraction Impacted f 1
1 hr 40 .4 2
.8 hrs 50 .5 3 .2 hrs 100 1
The fraction impacted f was calculated by dividing 100 customers into the number of customers impacted by the outage. The access availability of the router in this time period is
Access availability of router ()()()999.1000
1110002.18.5.14.1T D f 1N 1i i
i =?=×+×+×?=?
= =.
Example of IP Router Access Availability on a Port Basis The router has 200 access ports. In the time period, two outages occurred as indicated below
Outage
Duration D Access Ports Impacted Fraction Impacted f 1
1 hr 60 .3
2 2 hrs 20 .1
The fraction impacted f was calculated by dividing 200 access ports into the number of access ports impacted by the outage. The access availability of the router in this time period is
Access availability of router ()()9995.1000
5.1100021.13.1T D f 1N 1i i
i =?=×+×?=?
= =. Example of IP Router Access Availability on a Bandwidth Basis
The router has 5 access ports. In the time period, two outages occurred. Outage 1 had a duration of 2 hours and Outage 2 had a duration of 1 hour. The table below describes the router access ports and the impact of each outage.
Port Impacted x (1 = Yes, 0 = No) Access
Port
Bandwidth (Mbps) Outage 1 Outage 2 1
622 1 0 2 622 1 0 3
622 1 0 4
622 1 0 5 2488 0 1
The total access bandwidth for the router is
B = ==++++=P 1j j Mbps 49762488622622622622b
.
The fraction of the router impacted by Outage 1 is
f = ()()()()()5.4976
248806221622162216221B b x P
1j j
j =×+×+×+×+×= =.
The fraction of the router impacted by Outage 2 is f = ()()()()()5.4976
248816220622062206220B b x P
1j j
j =×+×+×+×+×= =. The table below displays these fractions along with the outage durations.
Outage
Duration D Fraction Impacted f 1
2 hrs .5 2
1 hr .5
The access availability of the router in this time period is Access availability of router ()()9985.1000
5.11100015.25.1T D f 1N 1i i
i =?=×+×?=?
= =. A.3 Examples of Average IP Router Access Availability for a Network
Three examples are given for calculating average IP router access availability for a network. Each example uses a different basis (customer, port, bandwidth) for calculating the weight given to each router’s access availability in the average.
Example of Average IP Router Access Availability for a Network (on a Customer Basis)
A network has three routers. In the time period under consideration, each router provides access service to C customers for in-service time T with access availability A as indicated in the table below. The total number of customer in-service hours for the network is
()()()000,000,1100050010002001000300T C
R 1k k k =×+×+×= =.
The average access availability of the network routers in the time period is
Average access availability = ()()()9992.9992.5.9995.2.9990.3.A w
R 1k k k =×+×+×= =.
Example of Average IP Router Access Availability for a Network (on a Port Basis)
A network has three routers. In the time period under consideration, each router provides service on P access ports for in-service time T with availability A as indicated in the table below. The total number of access port in-service hours for the network is
()()()000,000,110005001000400500200T P R
1k k
k =×+×+×= =.
The average access availability of the network routers in the time period is
Average access availability = ()()()9993.9994.5.9990.4.0000.11.A w
R 1k k k =×+×+×= =.
Example of Average IP Router Access Availability for a Network (on a Bandwidth Basis)
A network has four routers. In the time period under consideration, each router provides service on access ports with total bandwidth
B for in-service time T with availability A as indicated in the table below. The total Mbps in-service hours for the network is
()()()()000,440,121000248810004976100037325002488T B R
1k k
k =×+×+×+×= =.
The average access availability of the network routers in the time period is
Average access availability =
()()()()9994.9994.2.9997.4.9990.3.9994.1.A w
R 1k k k =×+×+×+×= =.
Appendix B: Techniques for Scaling Availability
Availability values may range from 0 to 1, but typically they are very close to 1. Common availability values such as .9999, .99999, and .999999 can be difficult to read and interpret. Several alternative expressions may be used to convey the information in availability values. Unavailability
Unavailability is the fraction of time that service is not available.
Unavailability = 1 – Availability
Downtime per Year
Downtime expresses unavailability in units of time, typically minutes per year.
Downtime per Year = Unavailability x 525,600 minutes