Is 50 ms Restoration Is 50 ms Restoration Necessary?Necessary?

Jeff Schallenberg

Optical Systems Planning, Teleglobe

jeff.schallenberg@teleglobe.com IEEE Bandwidth Management Workshop IX

Is 50 ms Restoration Necessary?Is 50 ms Restoration Necessary?

Outline: 1. Background

2. 50 milliseconds in the Standards

3. Laboratory Simulation of Brief Outages

4. Conclusions

Stentor Network Long (4500-km circumference) Matched-Node Ring*

The need to understand the REAL impact of brief outages

Is 50 ms Restoration Necessary? - BackgroundIs 50 ms Restoration Necessary? - Background

*”Field Deployment of Large SONET Ring Within Stentor (Canada) Telecom Network” – Sophie Legaultet al., NFOEC ‘98

•One large SONET ring proves in over several smaller ones:

•Fewer NEs and interfaces to buy

•Less fiber-km (no crosslinks)

•Theoretical availability is not significantly worse with longer fiber spans, even though there are more NEs and fiber in a single protection span

•BUT the BLSR protocol is slowed by the long propagation time to and fro around the large ring

•Matched Nodes using Drop-and-Continue on Protection also slow down intra-ring switching (like Extra Traffic)

Does operating outside the recommended 1200-km ring circumference limit threaten any of the services riding in the SONET frames?

Is 50 ms Restoration Necessary?- Is 50 ms Restoration Necessary?- 50 milliseconds in the Standards

Telcordia General Recommendations, especially GR-499 CORE “Transport Systems Generic Requirements - Common Requirements Criteria”

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

-PotentialVoicebandDisconnects(< 5%)-Affect CellReroutingProcess-TriggerChangeover ofCCS links

Outage Impac t

ProtectionSwitching Range

-Service"Hit"(Reframes) 1st

RestorationTargetRange

2ndRestorationTargetRange

-May DropVoicebandCallsDependingonChannelBankVintage

3rdRestorationTargetRange

-Call-Dropping(All CircuitSwitchedServices )-PL Disconnects-Potent ialPacket (X.25)Disconnects-Potential DataSessionTimeouts

4thRestorationTargetRange

-Packet(X.25)Disconnects-DataSessionTimeout

-NetworkCongestionExpectedSocial/BusinessImpact s

5thRestorationTargetRange

-PotentiallyFCCReportable

6thRestorationTargetRange

Outage Duration

50 ms 200 ms 2 s 10 s ~5 min ~30 min

Is 50 ms Restoration Necessary?- Is 50 ms Restoration Necessary?- 50 milliseconds in the Standards - continued

From T1.TR.68-2001: Examples of outage durations beyond which calls are abandoned, or other processes must be re-initiated (emphasis added):

SNA: programmable timeout 1.1 to 255 sX.25: idle state condition timer settable 1 to 30 sATM: cell rerouting process may be started within 200 ms, but could be longer in large or complex networksOld channel bank types (e.g., D1A, D1D, D2, D3) exhibit carrier group alarm times of 300 milliseconds to 500 milliseconds.Voice circuit: possible false disconnect at downstream switch trunk interface with 150 ms on-hook supervisory signal state

The TR further states “The time frame between 50 milliseconds to just under 200 milliseconds will have minimal impact to services. Any affected voiceband switched calls (voice or data) will have less than a 5% probability of being dropped.”

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - Laboratory Simulation of Brief Outages

1. The Telus Interruptor Switch – controlled “hits” with reed and optical switches

Switches, pots - Adjust rate, duration

Pushbutton - Start sequence

Pin jacks - n/o, n/c reed relays with 100-in series

Reed Relay for Coaxial (DS-3)

Optical Switch (OC-N)

+

6VDC

BNC Jacks for scope trigger

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - Laboratory Simulation of Brief Outages2.(a) Stentor Switching Lab – DMS-100 DS-0 SS7 Links

SSTP0 SSTP1

SDM1 SDM2

DS-1 Trunks

C-Links

A-Links A-LinksInterruptor Switch

•DS-0 hits <200 ms on a single link are not even detected

•200 ms < DS-0 hits < 1200ms on a single link cause “excess control codes” alarms at receive end, but alarms clear immediately

•TWO DS-0 HITS > 200 ms within 5 MINUTES on single link causes 60s “Delayed” state where the link is OOS - but the alternate link is OK

•To isolate a DMS (for 60s) requires BOTH links to be hit TWICE IN 5 MINUTES by >200 ms

Conclusion: SS7 links are immune to MULTIPLE brief (<200 ms) interruptionsand SINGLE moderate (>200 ms) hits

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - Laboratory Simulation of Brief Outages

2.(b) Stentor Switching Lab – DMS-100 DS-1 Trunks

SSTP0 SSTP1

SDM1 SDM2

DS-1 Trunks

C-Links

A-Links A-Links

Interruptor Switch

Conclusion: The DS-1 voice trunk network is immune to brief (<2s) hits

•Single and multiple hits <2s don’t affect a call. (Clicks and noise bursts are audible).

•Hits >2s cause DS-1 CGA, call is dropped, DS-1 trunk restores after 20s

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - Laboratory Simulation of Brief Outages

3. Stentor Data Lab – 56 kb/s data service in DS-1 signal

A 4.4-s “hit” onthe DS-1(actually 4, 1-s hits)

Is stretched by 20son the 56 kb/s data by the Tellabs 440MTE

Switch drive

Test set error output

Data Test Set MTE MTEDCS

1/0Interruptor Switch

DS-0 Loopback

56/64 kb/s DS-1 DS-1

Conclusion: DS-0-based data services are immune to moderate (<2.7s) DS-1 hits

•For DS-1 hits <2.7s, the 64 kb/s data stream recovers sync 20-30 ms after hit clears.•However when the DS-1 is hit for >2.7s, data sync is not restored for 15 -20s, depending on the particular MegaStreamTerminal Equipment (Newbridge and Tellabs were tested).

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - Laboratory Simulation of Brief Outages

4. Stentor Data Lab – NTSC Video Codec over ATM/OC-3

36150 36150OC-3X

CCD Camera

NTSCVideo

NTSCVideo

Colour Monitor

•If OC-3 connector opened/replaced within 5s (or so), the last frame is “frozen” by the codec until new data is received.

•If the OC-3 outage is longer than a few seconds, the codec inserts a colour-bar test pattern. When OC-3 is restored, video returns immediately.

Conclusion: Broadband services over ATM are immune to brief (<5s) hits and are not “shut down” by longish (>5s) interruptions

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - Laboratory Simulation of Brief Outages5. Stentor Access Lab – VF signal over Access SONET OC-12

DS-1 Loopback

CDS

Interruptor Switch

Lockout W

VF 1000Hz

RFT

FCOTStorageScope

A 12-ms hit recovers about 20ms after fiber restores.

>2s hit on fiber causes DS-1 CGA, signal is lost for 20s after fiber restores.

Conclusion:there is no drastic stretching of brief (<2s) fiber outages at tributary ratesdown to DS-0

5 ms/cm 10s/cm

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - Laboratory Simulation of Brief Outages6. Bell Platforms and Architectures Lab – DS-1 AAL1 over ATM and

SONET OC-48

PP1 PP2NB1 NB2

DS-3DS-3

OC-3

T

OC-12

OC-48

OC-48

DS-3

L A A

A

OC-12

A

A

A

OC-48

T

OC-3

DS-1AAL1Loopback

DS-1Test Set

DS-1 AAL1

Interruptor Switch

Lockout (W)

L

•Outages <2.5s restored immediately on ATM switches, causing NO alarms on ATM interfaces

•Outages >2.5s encountered the DS-1 CGA phenomenon, stretched for 34s, caused PAIS alarms on Newbridge 2, RFI on Passport 1

Conclusion: The ATM Core and Edge networks are immune to brief (<2.5s) hits

(Edge Switch)

(Core Switch)

(Edge Switch)

Is 50 ms Restoration Necessary? - Is 50 ms Restoration Necessary? - . Conclusions

Today’s network is quite resistant to brief outages

200 ms outages would not jeopardize services (data, video, voice)

equipment design could be simplified

equipment costs could be reduced

No, not really!