SONET.pdf

CSC/ECE 791B – Survivable NetworksSONET Protection Switching

George N. Rouskas

Department of Computer Science

North Carolina State University

CSC/ECE 791B, Spring 2008: SONET Protection Switching Copyright c© 2007 by George N. Rouskas – p.1

http://www.csc.ncsu.edu/faculty/rouskas/

Outline

1. Protection Switching Architectures

2. SONET Ring Types

3. Two-Fiber Protection

4. Four-Fiber Protection

5. Automatic Switching Protocol (APS)


Protection Switching

Network must continue to provide reliable serviceseven in the presence of

failures

errors

poor signal quality

Protection techniques:

ensure survivability

involve the provision of redundant capacity

reroute traffic when failures occur


APS Protocol

OAM protocol detects abnormal conditions

Automatic protection switching (APS) protocol:

switches traffic from working to protection entity upon failure

no manual intervention

Manual intervention necessary for repairing failed entity

Revertive or non-revertive operation

ATM protection techniques and APS protocol very similar


Non-Ring APS

Four architectures:

1. 1+1 switching

2. 1:1 switching

3. 1:n switching

4. m:n switching

T-carrier employed protection switching


1+1 Switching

Source Sink

Working path

Protection path

Bridge

Protection Domain

Selector


1:1 Switching

Source Sink

Working path

Protection path

Protection Domain

SelectorBridge


Extra Traffic Capability

Extra traffic:

low priority, not protected traffic

occupies protection entity under normal operation (no failures)

preempted (dumped) when working entity fails and protectedtraffic switched over to protection entity

sold at deep discount


1:n Switching

Source 1

Sink n

Sink 2

Sink 1

Working path

Switch

Switch SwitchProtection path

Source 2

Source n

Protection Domain

Selector


Protection in the Access Network

CO

Business

CO

Business

Same feeder,Diverse conduit

Same feeder,Diverse sheath

CO

Business

Different feeders,Diverse route

CO CO

Business

Different feeders,Different COs


SONET Rings

Self-healing rings:

services automatically restored following a failure or signaldegradation

restoration times less than 60 ms

Deploy fiber for loop diversity:

1. separate fiber sheath

2. separate conduits

3. route diversity: take different physical routes from src to dest


SONET Ring Types

Attribute Choices

Number of fibers per link 2-fiber

4-fiber

Direction of the signal Unidirectional

Bidirectional

Level of protection switching Line switching

Path switching


Unidirectional Rings

Only one direction around the ring used for two-way communication

→ Asymmetric delays

All working traffic travels in clockwise direction

Opposite direction used for protection


Unidirectional Rings (cont’d)

Span #7

Span #3

Span #1

Span # 5

Span

#4 Span #8 Sp

an #

6 Span #2

NE 2

NE 3

NE 1

NE 4


Bidirectional Rings

Physically indistinguishable from unidirectional rings;difference is in direction of traffic flow

Under normal routing, both directions of a connection:

travel along ring through same ring nodes

travel in two opposite directions

→ Symmetric delays

Working traffic in both clockwise and counter-clockwise direction

If links between NE1-NE2 fail, protection switching uses spansbetween NE2-NE3, NE3-NE4, and NE4-NE1


Bidirectional Rings (cont’d)

Span #3

Span

#4 Span #8 Sp

an #

6 Span #2

NE 2

NE 3

NE 1

NE 4

Span #1

Span # 5

Span #7


2-Fiber vs. 4-Fiber Rings

2 or 4 fibers between each pair of SONET nodes in the ring

2-fiber rings

robust enough for small geographical area (within city)

may survive single failure, will partition with two or more

4-fiber rings

used for regional, national backbones

may survive multiple failures

4 fiber unidirectional rings: quite uncommon

2-fiber vs. 4-fiber bidirectional rings


2-Fiber Bidirectional Ring

NE1 NE2Span #1 Working channels

Span #1 Protection channels

Span #5 Working channels

Span #5 Protection channels

Each fiber span carries both working-traffic channels and protectionchannels

At most half the channels on each fiber can carry working traffic


4-Fiber Bidirectional Ring

NE1 NE2Span 1A Working channels

Span 1B Protection channels

Span 5A Working channels

Span 5B Protection channels

Working and protection pairs carried over different fibers

Twice as much fiber cable, but each fiber can be used to capacity


Path Switching vs. Line Switching

Concepts valid in general mesh networks (not just rings)

Path switching:

restoration of traffic handled by source/destination of each affectedtraffic stream

source/destination reroute traffic in the event of a failuresomewhere in the route

affected traffic streams may take different protection routes

also called path protection

implemented in a 1+1 or 1:n arrangements


Path Switching vs. Line Switching (cont’d)

Line switching: restoration of traffic is handled by the nodes at theends of failed link, not the sources/destinations

Two ways to implement:

1. span protection: if a fiber is cut between two nodes, traffic isswitched to another fiber between same two nodes

2. line protection: traffic is switched to another route through thenetwork between the same two nodes

all affected traffic streams take same protection route


Path Switching vs. Line Switching (cont’d)

Connection

(a) Normal operation (b) Path switched restoration (path protection)

(c) Span protection, a form of line switching (d) Line protection, another form of line switching


Common SONET Ring Types

All 8 ring types are possible, but three have become common:

1. UPSR: two-fiber unidirectional path-switched rings

2. BLSR/2: two-fiber bidirectional line-switched rings

3. BLSR/4: four-fiber bidirectional line-switched rings


UPSR

1+1 protection: traffic from A to B sent simultaneously onworking/protection fibers

B monitors both fibers, selects the better signal

Fast restoration:

action required only at receivers

no need for complicated signaling (APS) protocol

But: asymmetric delays

not a problem for voice traffic

problem for TCP window flow control


UPSR (cont’d)

NE 2NE 1

NE 4 NE 3

Protection fiberProtection trafficWorking traffic Working fiber


UPSR (cont’d)

No spatial reuse:

a bidirectional connection uses capacity on each link of ring

max traffic on ring equal to link speed

No limit on number of nodes, length of ring

Simple, easy to implement, low cost

Popular in lower-speed local exchange and access networks


BLSR/4

Two fibers for working traffic, two fibers for protection

Working traffic carried on both directions along the ring

Traffic routed on shortest path between end nodes

Spatial reuse:

each connection uses capacity only on shortest path

aggregate traffic can significantly exceed link speed

shortest path routing maximizes spatial reuse

Extra traffic capability (1:1 protection)


BLSR/4 (cont’d)

NE 1

NE 4 NE 3

NE 2

Working fiber Protection fiber

Working traffic


BLSR/4 Protection Mechanisms

Span protection: traffic switched to protection fiber between two nodeswhere failure occurred

transmitter/receiver failures on a working fiber

working fiber cuts

Line protection: traffic rerouted around the ring on protection fibers

cuts of both protection and working fibers along a link

node failures


BLSR/4 – Normal Operation

NE 4

NE 3NE 2NE 1

NE 5NE 6

Working

Protection


BLSR/4 – Span Protection

NE 4

NE 3NE 2NE 1

NE 5NE 6

Working

Protection


BLSR/4 – Line Protection

NE 5NE 6

Working

Protection

NE 4

NE 3NE 2NE 1


BLSR/2

Protection fibers embedded within working fibers

Both fibers used to carry working traffic

Half the capacity on each fiber reserved for protection

Span protection not possible

Line protection similar to BLSR/4:

upon link failure, traffic rerouted along other part of ring usingprotection capacity on two fibers

traffic mapping a tricky problem

extra traffic capability (1:1 protection)


BLSR/2 – Normal Operation

OC-12 BLSR/2 with 12 STS-1s

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

NE 1 NE 2 NE 3

NE 4NE 5

STS-1 #3STS-1 #3


BLSR/2 – Line Switching

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

W: 1-6P: 7-12

NE 1 NE 2 NE 3

NE 4NE 5

STS-1 #3STS-1 #3


BLSRs

More efficient than UPSRs for distributed traffic patterns

Protection capacity shared among all connections

Example: 10-node ring, 1.5 Mbps between adjacent nodes

UPSR requires 15 Mbps protection capacity on each fiber

BLSR/2 requires 1.5 Mbps protection capacity on each fiber


BLSRs (cont’d)

Maximum number of nodes: 16

Maximum ring length: 1200 Km (6 ms propagation delay)

BLSRs deployed in regional/national high-speed (OC-48, OC-192)networks

BLSR/4 can handle more failures than BLSR/2


SONET Matched Nodes

ADM

ADM

ADM

ADM

ADM

ADM

ADM

ADM

MN

MN MN

MN

1+1 protection

SONET Ring #1 SONET Ring #2


Automatic Protection Switching (APS)

APS takes place at the SONET line level

Very complex task

ANSI APS document (T1.105.01-998) 100 pages long!

Only basic operation explained here

Emphasis on role of K1, K2 bytes of LOH

ATM APS protocol similar, K1, K2 bytes in APS ATM cells


APS Objective

Whether traffic is received over the working or protection fiberis determined by:

1. the status of the bridge at source node

2. the status of the selector at destination node

Objective: establish agreement between source and destinationregarding the status of bridge/selector

K1, K2 LOH bytes used by APS protocol for this purpose


APS Events

Protection switching: a change in the current position of thebridge/selector

Initiated due to certain events:

1. externally initiated commands, e.g., forced switch, manual switch,lockout of protection, etc.

2. automatically initiated command, e.g., loss of signal (LOS), loss offrame (LOF), signal degrade (due to parity errors), etc.


APS Protocol: K1, K2 bytes

K1 byte:

switch request (protection switching event) – 4 bits

destination node – 4 bits→ max 16 SONET nodes

K2 byte:

source node – 4 bits

long/short bit

status (of bridge/selector) – 3 bits

Source/destination use K1, K2 bytes to coordinate protectionswitching actions


APS Protocol Operation

Each node:

uses local priority logic to rank (possibly many) local events

encodes highest priority event E1 into K1 byte to be sent

extracts event E2 last received by remote entity

uses global priority logic to rank events E1, E2

let E be the highest priority event among E1, E2:

sets the status of local bridge/selector based on E

encodes status in the K2 byte to be sent

if status 6= status of last K2 byte received, mismatch alarm


WDM SONET Rings

1. W point-to-point rings, each on one of W wavelengths

high cost: W OADMs, SONET ADMs at each node

independent rings

severe electro-optic bottleneck

2. Static virtual topology, based on traffic pattern

fewer OADMs, SONET ADMs, alleviates bottleneck

3. Dynamic virtual topology

requires sophisticated OXCs, traffic grooming capabilities


SONET.pdf

Documents

protection cscece

protection entity

fiber protection

protection switchingnetwork

copyright c

path protection domaincscece

signal level of protection

ber bidirectional rings