Convergent Communications for the Olympics: Netw ork Reliability - - PowerPoint PPT Presentation

Convergent Communications for the Olympics:

Netw ork Reliability Challenges

Spilios E. Makris, Ph.D.

Director, Olympic Program Network Reliability & Risk Services

Prepared for: WOCC 2007 NJIT Newark, NJ April 27-28, 2007

WOCC 2007 – 2

The Issue

Growing concern from the telecom community about the reliability/availability of convergent (IP- based) telecommunications networks, including the services provided under failure conditions Growing concern from the telecom community about the reliability/availability of convergent (IP- based) telecommunications networks, including the services provided under failure conditions

WOCC 2007 – 3

The Games Netw ork Challenge

How to design, build, and operate a world-class network for the Olympic Games, and convince the world that the infrastructure is capable of meeting the needs of the IOC, the Organizing Committee, the National Olympic Committees, government agencies, sponsors, broadcasters, and spectators

“ “Perfect Perfect Games Network Games Network” ”

WOCC 2007 – 4

User-to-User Connection on an IP Netw ork

(ITU-T Rec. Y.1541)

TE TE GW

. . . . . .

Network Section End-to-End Network (Bearer Service QoS) Network Section Network Section Customer Installation Customer Installation User-to-User Connection (Teleservice QoS) TE GW Terminal Equipment GateWay Protocol Stack LAN LAN

IP Network Cloud

NI NI NI Network Interface GW GW GW GW GW LAN Local Area Network

WOCC 2007 – 5

Role of IP Nodes in a Netw ork Section

R Router R R R R R R Access Distribution Core GW GW GateWay GW GW

WOCC 2007 – 6

Reliability/Availability Framew ork for IP-based Netw orks & Services

Address:

– The growing concerns about the reliability/availability of IP networks – Service quality/performance under failure conditions

Define a set of metrics and guides to promote consistent:

– SLAs that are rich in reliability/availability attributes – Network element reliability/availability/maintainability requirements

Discuss:

– Impact of failures and operational activities (network dimensioning,

traffic engineering, & capacity management) on service reliability

– QoS benchmarks to define failure thresholds

WOCC 2007 – 7

Framew ork Overview

Service User Service Provider

Services View

Ntwk Architecture Network Elements

Network View

Operations User User Access Info Transfer Disengagement Congestion Failure

Option A Option B (ITU-T Rec. I.350)

…

Option N Applications Services Service Infrastructure Transport Infrastructure

Services View Network View User-perceived Performance Metrics Network Performance Metrics Service Level Agreements Network Requirements Framework

WOCC 2007 – 8

Framew ork Application

Services View Network View Observable

• Service Denial
• Subscriber Downtime
• etc.

F a i l u r e s Congestion Environmental Incidents

Measurable

• Impact
• Duration
• Frequency

Next Generation Networks (IP-based) Bottom-up

• Operations: measure

‘defective events’ for corrective actions

Top-down-bottom-up

• Design: SLA-driven

network design & engineering

• Validation: via fault

insertion for service impact

Consistency across the industry

Framework

WOCC 2007 – 9

Important Considerations for IP-based Services

Applications provided by IP-based service offerings experience

network outages as:

– Packet delay – Packet loss

Network outage impacts differ depending on:

– The type of service – Extent & duration of outage

User tolerance of service impairments depends on:

– Service type – Service criticality – His/her willingness to pay for the service

The reliability/availability requirements may vary for multi-service, IP-based networks

WOCC 2007 – 10

Measurement

Service User Metrics Network Metrics Business Business Drivers Drivers

Field Reliability/ Availability Performance

Measured Failure Modes

• Impact
• Duration
• Time to failure
• Population

Metrics Equations

Network Architecture Reliability/Availability SLA

WOCC 2007 – 11

Reliability/Availability-SLA Process Steps

Propose Initial Metrics Assess Risk R/A-SLA Reqs Set R/A-SLA Legacy Expectations Regulatory Requirements Service Criticality Market Segment Competition Business Goals

R/A-SLA Ntwk Solution Requirements

Set Network Solution Reqs

R/A-SLA Template

Propose Initial R/A-SLA reqs

WOCC 2007 – 12

Example SLA Guide for an IP-based Service

Service:

VoIP/Data Service Access for Large Enterprise

SLA Requirements

Service Activation < 24 hours Service Restoration < 30 min for 100% of incidents Subscriber downtime < 10 min/yr for outages >30 sec averaged over one year Catastrophic downtime < 0 min/yr for outages >30 sec (> 90% of users) averaged over one year Major downtime < 2 min/yr for outages >30 sec (10-90% of users) averaged over one year Minor downtime < 8 min/yr for outages >30 sec (<10% of users) averaged over one year

Service:

VoIP/Data Service Access for Large Enterprise

SLA Requirements

Service activation < 24 hours Service restoration < 30 min for 100% of incidents Subscriber downtime < 10 min/yr for outages >30 sec averaged over one year Catastrophic downtime < 0 min/yr for outages >30 sec (> 90% of users) averaged over one year Major downtime < 2 min/yr for outages >30 sec (10-90% of users) averaged over one year Minor downtime < 8 min/yr for outages >30 sec (<10% of users) averaged over one year

Service User Metrics Network Metrics Business Business Drivers Drivers

Business Technology

WOCC 2007 – 13

Case Study: Olympic Games

At Stake: “Perfect” Olympic Network for voice, data, and video Issue: Whether the future IP-based Olympic Networks will be as robust as the “traditional” TDM-based ones Migration path: From overlaid video, data, voice Olympic Networks, to “mixed” Olympic/PSTN, to IP-based Olympic Ntwks Action: Ensure the IOC, broadcasters, and other stakeholders that Olympic telecom services will not be degraded regardless of the underlying technology Resolution: Benchmark IP-based Olympic Networks and offer reliability/availability SLAs

WOCC 2007 – 14

Athens 2004 Olympic Netw ork Results: Service Availability (Source: OTE)

Video:

– Multilaterals: 100% – Unilaterals: 100%

WAN Games and Admin: 100% Commentary circuits:

– 4-wire circuits: 100% – 2 Mb/s circuits: 100% – WAN high-quality circuits: 100%

National & International TV circuits: 100% CATV from IBC to TERs (Telecom. Equip. Rooms): 100% Voice services (PSTN, Olympic VPN, ISDN-BRA, ISDN-PRA): 99.997% Card-phones: 99.9994% 2 Mb/s Leased lines: 99.99% ADSL: 99.95%

WOCC 2007 – 15

Industry Challenges for IP-based Netw orks & Services

Case: 2010 & Beyond Olympic Netw orks

Lack of industry consistency for reliability analysis or benchmarks of

IP-based networks

Need for network providers to know if what they are building will deliver

the service reliability/performance required by end users

Service providers specify service availability within own network

domains; end-user services delivered across multiple domains are virtually impossible to guarantee

Best-in-class reliability/availability SLAs while minimizing the cost of

• perations and maintenance

Coordination of efforts in different industry forums (ATIS, ITU-T, IETF,

etc.) and government-sponsored activities (e.g., NRIC)

WOCC 2007 – 16

To Fulfill the Promise of Service Convergence in IP-based Netw orks

The industry needs a set of reliability metrics and terminology that are common across the network for which individual service availability requirements can be specified The industry needs a set of reliability metrics and terminology that are common across the network for which individual service availability requirements can be specified

WOCC 2007 – 17

ATIS (PRQC) Standards & Technical Reports (TRs) on Netw ork Reliability/Survivability

Published:

“Enhanced Network Survivability Performance”, T1.TR.68-2001,

February 2001

“A Reliability/Availability Framework for IP-based Networks and

Services”, T1.TR.70-2001, September 2001

“Access Availability of Routers in IP-based Networks” T1.TR.78-

2003, January 2003 Expected:

“Standard on End-to End Service Availability” (expected in 2Q2007) “Standard on Reliability Related Metrics and Terminology for Network

Solutions” (expected in 4Q2007)