Optical layer protection approaches Pter Babarczi - - PDF document

SLIDE 1

1

Optical layer protection approaches

Péter Babarczi babarczi@tmit.bme.hu

(János Tapolcai tapolcai@tmit.bme.hu)

2

1+1 Protection (disjiont pair of paths)

• 200km lightpath 0.99064

) 1 ( 1

1 i m i

A A − ∏ − =

=

53 min/year

• utage

Parallel rule: A = 0.99979

3

Disjiont working and protection paths – Why?

• The availability of each link is 0.999

Working path Protection path Working path Protection path

0.999994 0.99899

Source Source Destination Destination

4

Availability improvement techniques

• Duplicate network devices

– Redundant optical switches – “dual homing”

• Two access link per user
• Often this is not enough
• Conscious solution:

– Protection

• Network recovery
• Resilience schemes

Internet protocol stack

• application: supporting network

applications – FTP, SMTP, HTTP

• transport: host-host data transfer

– TCP, UDP

• network: routing of datagrams from

source to destination – IP, routing protocols

• link: data transfer between

neighboring network elements – PPP, Ethernet, SDH

• physical: bits “on the wire”

– DWDM

application transport network link physical

Topologies

Acces Metro Backbone/ core

Star or ring topology Ring topology Previously: (Virtual) Ring topology Now: Mesh topology

SLIDE 2

message segment datagram frame

source

application transport network link physical

Ht Hn Hl M Ht Hn M Ht M M

destination

application transport network link physical

Ht Hn Hl M Ht Hn M Ht M M

network link physical link physical

Ht Hn Hl M Ht Hn M Ht Hn Hl M Ht Hn M Ht Hn Hl M Ht Hn Hl M

router switch

Encapsulation

Which layer should we use to reach higher availability?

8

Layer 5 - Application layer

• Supporting network

applications

– Process-to-process communication using sockets (by exchanging messages) – Berkeley Socket is a widely used API (Application Programming Interface)

• socket() – creates an

endpoint for communication

• connect() – connects a

socket

• listen() - waiting for

incoming connection

process TCP with buffers, variables socket host or server process TCP with buffers, variables socket host or server Internet controlled by app developer

9

Layer 4 – Transport Layer

• host-host data transfer
• Responsible for sending data between end

nodes

– Transmission Control Protocol (TCP)

• Reliable end-to-end connection
• TCP sent in IP packets

– Error correction, congestion control, flow control

– User Datagram Protocol (UDP)

• Unreliable, but fast

10

Layer 3 – Network Layer

• Routing of datagrams from source to destination
• Transport segment from sending to receiving

host

– Efficiently

• As much packets as possible
• Best effort

– Self adaptive

• to the changing topology fast (link state protocols like OSPF)

– Fully distributed manner

• 1981: “end-to-end principle”

– No intelligence in network core, only on the edges

11

Layer 2 – Data link layer

• Data transfer between neighboring network

elements

– has responsibility of transferring datagram from one node to adjacent node over a link – Aggregated traffic

• State-dependent
• Finer routing
• Dynamically configured

– Traditionally network are statically configured – Reserve flows on the links

• Reliable infrastructure

– Error detection, error correction, flow control

12

Layer 1 – Physical layer

• Optical channels in the core/backbone

– DWDM – Dense wavelength division multiplexing – Deploy lightpaths between users

SLIDE 3

13

Optical layer protection

• Protect connections in lower layers (1-2)

– Waiting for upper layer protocols (e.g. OSPF in layer 3) to adapt to the failure introduce high latency in the network

• A network recovery mechanism, which

– Detects the failure – Reroute the traffic on an operating path

Working path Protection path Swithing or splitting node Switching node Recovered segment Sorce Destination

14

Recovery in multi-layer networks

• Rapid reaction in the upper layer cause redundant

recovery action

Link failure The link is protected by the optical layer Refresh the routing table (link is failed) Refresh the routing table (link is operating) Detect, that the link is

• perational again

100 ms 10 sec 10 sec ALARM Traffic

Source: RHK

15

Recovery cycle of a protection method

Recovery time Recovery operation (switching) time Fault notification time Fault detection time time notification The service is

• perational

Failure detected by the nearast node The protection path is deployed Data flow arrives at the destination node

failure

Hold-Off time Sending fault notification The service is

• perational

After failure is repaired: reversion cycle to optimize the usage of network resources

Traffic Recovery time 16

Protection and restoration

100%, fast No guarantee, slower pre-planned (protection) after failure event occures (restoration) link path segment link path segment dedicated shared dedicated shared dedicated shared Failure dependent Faiure independent (the faied element is unknown) Failure dependent Faiure independent (the faied element is unknown)

Different protection approaches from down to top (e.g. Dedicated Path protection or Failure Dependent Shared Link Protection) 17

Link, segment and path protection

1 2 3 4 7 5 8 9 6 1 2 3 4 7 5 8 9 6

Link

1 2 3 4 7 5 8 9 6

Path

Failure

1 2 3 4 7 5 8 9 6

Segment

Failure Failure Back hauling

18

Dedicated protection

The working and protection paths

• perate during

the whole holding time of the connection (hot stand-by)

Protection resources are additional

1 2 1 1 1 1 1

SLIDE 4

19

Backup resources are not additional, their maximum need to be reserved

Shared protection for single link failures

• If two working

paths are link disjoint, their protection capacity is shareable.

Shared bandwidth

1 1 1 1 1 1 1

20

Dedicated 1+1 protection

• Signal is sent parallel on two disjoint paths

– Both configured and signaled at connection setup

• If the destination node sense the degradation of the

signal on the working path, switches over the protection path (single end switching)

• Simple (thus widely deployed) but capacity consuming

(hot stand-by)

• Recovery time:

Path 1 Path 2 Source 1+1 Destination

Switching time Fault notification time Fault Detection Time time

failure

Hold-Off Time

Splitter Switch

21

Dedicated 1:1 protection

• Two disjoint paths are reserved

– Data is sent only on the working path – Protection resources can be used for route low priority traffic

• If the working path failes, switching is needed both in

source and destinaiton node

– Low priotity traffic is preempted

• Recovery time:

Working path Protection Path 1:1 Source Destination

Switching time

Fault notification time Fault Detection Time

time

failure

Hold-Off Time

Switch Switch

22

Dedicated m:n protection

• Common pool m of protection paths are given

for a set of working paths n

• Consumes less resources
• Lower availability
• Calculate the availabilty of 1:1 protection:

– Aw, Ap

• And for 1:2 protection:

– Aw1, Aw2, Ap

Destination 1:n ... n working paths Source protection path

A=1-(1-Aw)(1-Ap)=Aw+Ap-AwAp A=Aw1Aw2+(1-Aw1)Aw2Ap+Aw1(1-Aw2)Ap

23

Dedicated m:n protection

m:n ... ... Source Destination m protection paths n working paths

Topologies

Acces Metro Backbone/ core

Star or ring topology Ring topology Previously: (Virtual) Ring topology Now: Mesh topology

SLIDE 5

25

Mesh topology

• Disjoint working and protection paths

– The problem is more complex than in ring topology

Working path Protection path 26

Shared Risk (Link) Group

• Set of network elements which share a

common risk of failure (and fail together)

– physical devices – protocols – links – nodes

SRLGs Assumption: SRLG failures are independent!

27

SRG disjoint pair of paths

• Given:

– Graph representation of the topology G=(V,E) – list of SRGs,

• Objective:

– Find SRG disjoint working and protection path – SRGs result an s-d cut in the topology are not considered in the routing problem

Védelmi út 28

Suurballe’s algorithm – finding link disjoint paths (1)

• Shortest path finding algorithm used (e.g.

Dijkstra’s algorithm)

Src Dst 1 1 1 2 2 1 1

29

Suurballe’s algorithm – finding link disjoint paths (2)

• Direct edges in the opposite direction and

find shortest path in the modified graph

Src Dst

• 1

1

• 1

2 2 1

• 1

30

Suurballe’s algorithm – finding link disjoint paths (3)

• Get the pair of disjoint paths (first the red
• ne)

Src Dst

Use the other shortest path if needed

SLIDE 6

31

Suurballe’s algorithm – finding link disjoint paths (4)

• Get the pair of disjoint paths (the blue one)

Src Dst Use the other shortest path if needed

32

Suurballe’s algorithm – finding link disjoint paths (5)

• Link-disjoint pair of paths

Src Dst 1 1 1 2 2 1 1

33

Suurballe’s algorithm – finding node disjoint paths (1)

• Link disjoint paths are not necessary node disjoint

– How can we get node disjoint pair of paths with Suurballe’s algorithm?

Src Dst

34

Suurballe’s algorithm – finding node disjoint paths (2)

• Use the node splitting technique:

Src Dst

The previous solution is not link-disjoint in this representation!

35

Finding SRG disjoint pair of paths

• The problem is NP-hard in general
• In our case single link (and mostly single

node) failures are protected

– Polynomial time algorithms exist

• finding link disjoint paths
• finding node disjoint paths
• The sum of the pair of disjoint paths is minimal

among all disjoint path-pairs in the network

• Trap avoidance property

Flow problem, Suurballe’s algorithm Node splitting and flow problem

36

References

• Dr. Chidung LAC, “Telecommunication

network reliability”

• D. Arci, et.al, “Availability models for

protection techniques in WDM networks”

• Computer Networking: A Top Down

Approach Featuring the Internet, 3rd

• edition. Jim Kurose, Keith Ross Addison-

Wesley, July 2004.

• J. Vasseur, M. Pickavet, and P. Demeester.

Network recovery: Protection and Restoration of Optical, SONET-SDH, IP, and MPLS. Morgan Kaufmann Publishers, 2004. Computer Networking: A Top Down Approach Featuring the Internet, 3rd edition. Jim Kurose, Keith Ross Addison-Wesley, July 2004.