Content .. 3 4 5 6 ..

OptiX OSN 7500 Intelligent Optical Switching System. Technical Manual - part 5

Company A

Data

NE1

NE2

NE3

NE4

Corp A

Port A

Port B

Tunnel label

switching

Company B

L2 MPLS network composed

by OptiX OSN Products

Company

Tunnel Label

VC Label

Data

NE4 strips the MPLS lables

and Transfers the frame to

corresponding ports

Figure 6-3 The EVPL service with MPLS label

4. EPLAN Service

The OptiX OSN 7500 supports Layer 2 switching of Ethernet data, i.e. the EPLAN

service, which can be transferred according to their destination media access control

(MAC) addresses.

As shown in Figure 6-4, respective LANs of Company A and B are connected to four

NEs. The Ethernet service between the NEs is not of a fixed point-to-point type. For

example, a user of Company A connecting to NE3 may want to communicate with

users of Company A connecting to other three NEs. That is, the flow direction of

services is not definite. The Ethernet Layer 2 switching function provided by the OptiX

OSN 7500 can be employed to solve such a problem. For example, an Ethernet MAC

address transfer table will be formed in the system when the relevant settings are

made to NE3. The system can learn to periodically update the table. Then, the data of

Company A and B accessed at NE3 will be transmitted to their destinations over

different VCTRUNKs selected according to their MAC transfer table or over the same

VCTRUNK.

In this way, the system configuration is significantly simplified and the bandwidth utility

is improved. In addition, the corresponding maintenance and management becomes

convenient for the operator.

6-5

Company B

NE1

NE2

NE4

Company A

SHR

NE3

Company B

MAC Address

Destinati

VC-Trunk

Traffic flow

MAC 1

NE1

①

MAC 2

NE4

②

Company A Company B

MAC 3

NE2

③

…

Figure 6-4 Layer 2 switching of Ethernet service

5. EVPLAN Service

The OptiX OSN 7500 adopts the Martini MPLS Layer 2 VPN encapsulation format to

support the Ethernet virtual private LAN (EVPLAN) service.

EVPLAN service implements the multipoint-to-multipoint connection of user sites.

Users regard the EVPLAN network as a big VLAN where the user service can be

converged. As shown in Figure 6-5, when the user’s Ethernet frame (the source

address is MAC H, and the destination address is MAC A, B or C) enters the PE

equipment, the system will search the Layer 2 transfer table for the internal label (VC

label). Then, the frame is transferred to the corresponding tunnel, where it is attached

with the external label (tunnel label). Thus, different LSPs are set up according to

different addresses. The MPLS labels are switched at the LSP. And then transferred to

the corresponding PE equipment, where the tunnel and VC labels are striped. After

that, the Ethernet frame is transferred to the corresponding output port according to

the Layer 2 MAC transfer table.

6-6

Address =

Core

MAC C

Branch C

Address =

MAC B

MAC A

Branch A

Branch B

LSP3

LSP1

LSP2

Transferd to

corresponding

port via the Layer

2 route table

Address =

Source

Sink

Tunnel Label

VC Label

LSP

MAC H

MAC A

LSP1

MAC H

MAC B

LSP2

Headquarters

MAC H

MAC C

LSP3

Figure 6-5 Application of EVPLAN service

6.1.3 Protection

The Ethernet service of the OptiX OSN equipment takes the protection of several

levels, including:

Protection of the spanning tree, LCAS and flow control

Protection of optical transmission layer, such as MSP and SNCP

1. LCAS

LCAS provides an error tolerance mechanism, enhancing the reliability of

concatenation. It has the following functions:

Configure the system capacity, add or reduce the number of VC involved in the

concatenation and change the service bandwidth dynamically without damaging

the service.

Protect and recover failed members.

As shown in Figure 6-6, LCAS can add or delete members to increase or decrease

the bandwidth dynamically without affecting the service.

6-7

I want another 10 M

bandwidth.

Member

Headquarters

Branch

Member

Headquarters

Branch

New member

MSTP

Figure 6-6 LCAS adjusts bandwidth dynamically

As shown in Figure 6-7, LCAS can protect the Ethernet service. When some

members fail, the failed members will be deleted automatically. While other members

remain transmitting data normally. When the failed members are available again, they

will be recovered automatically, and the data will be loaded to them again.

MSTP network

Member

Headquarters

Branch

Failed member

Member

Headquarters

Branch

Delete failed member

MSTP

Figure 6-7 LCAS protects the concatenation group

6-8

The Ethernet boards support spanning tree protocol (STP) and rapid spanning tree

protocol (RSTP). When STP is started, modify the logic network topology to avoid

potential broadcast storm.

3. Flow Control

The Ethernet interface supports IEEE 802.3X flow control, minimizing the packet loss

caused by congestion. As shown in Figure 6-8, the node connecting with Ethernet in

duplex mode sends PAUSE frame to ask the receiving node to stop transmitting frame

signals within a pause-time (N seconds), so as to avoid frame loss.

Ethernet switch

Duplex

Data transmission

Buffer is not full.

Data transmission

Buffer is full.

Pause frame

Buffer is full.

Pause-time = N seconds

Pause frame

Buffer is not full.

Pause-time = 0 second

Data transmission

Buffer is not full.

Ethernet switch

MSTP

Figure 6-8 Flow control at the Ethernet side

6.2 RPR Features

This section introduces the RPR features of the OptiX OSN 7500 in terms of function,

application and protection.

6.2.1 Function

The EMR0 and EGR2 boards of the OptiX OSN 7500 supports resilient packet ring

(RPR) features defined by IEEE 802.17. RPR employs a dual-ring structure utilizing a

pair of unidirectional counter-rotating rings, as shown in Figure 6-9. Both the outer ring

and the inner ring bear data packets and control packets, featuring high bandwidth

utilization. The control packets on the inner ring carry control information for the data

packets on the outer ring, and the control packets on the outer ring carry control

information for the data packets on the inner ring. The two rings act as backup and

6-9

Node 1

Outer ring data

Outer ring control

Node 2

Node 4

2.5 Gbit/s RPR

Inner ring control

Inner ring data

Node 3

Figure 6-9 RPR ring

1. Basic Functions

The EMR0 and EGR2 boards of the OptiX OSN 7500 support the resilient packet ring

defined by the IEEE 802.17 standard. Table 6-2 lists their functions.

Table 6-2 Functions of Ethernet boards supporting RPR

Board

EMR0

EGR2

Feature

Interface

1xGE+12xFE

2xGE

Service frame format

MPLS

JUMBO frame

Supported, 9600 bytes

Max. uplink bandwidth

2.5 Gbit/s

RPR bandwidth

VC-3, VC-3-2v, VC-4,

VC-4-Xv (X≤16)

Ethernet virtual private

Supported

line (EVPL)

Ethernet virtual private

Supported

LAN (EVPLAN)

MPLS

Supporting MPLS frame format, constructing EVPL service

and EVPLAN service

VLAN

6-10

EMR0

EGR2

Feature

MAC address table

64k

MPLS label

Rapid spanning tree

Support STP and RSTP

protocol (RSTP)

Multicast (IGMP

Supported

Snooping)

RPR protection

Steering, Wrapping,

Wrapping + Steering

RPR protection period

<50ms

Encapsulation format

GFP-F, compliant with ITU-T

GFP-F, compliant with

G.7041

ITU-T G.7041

LAPS, compliant with ITU-T

X.86

LCAS

Supported, compliant with

ITU-T G.7042

CAR

Based on port or port + VLAN,

Based on port or port +

with the granularity as 64

VLAN, with the granularity

kbit/s

as 64 kbit/s

Flow control

Supported, compliant with

IEEE 802.3X and IEEE

802.3Z

Weighted fairness

Supported

algorithm

Topology discovery

Supported

Service class

A0, A1, B_EIR, B_CIR and C

A0, A1, B_EIR, B_CIR and

2. Service Class

The user service has three classes, A, B and C. Class A falls into A0 and A1. Class B

falls into B_CIR (Committed Information Rate) and B_EIR (Excess Information Rate).

Table 6-3 gives the difference of these classes.

6-11

Class

Sub-class

Bandwidth

Jitter

Fair algorithm

Application

Allocated,

Low

Irrelevant

Real time

irreclaimable

Allocated,

Low

Irrelevant

Real time

reclaimable

B_CIR

Allocated,

medium

Irrelevant

Near real time

reclaimable

B_EIR

Opportunistic

High

Relevant

Near real time

Opportunistic

High

Relevant

Best effort

3. Topology Discovery

The topology discovery function realizes the plug and play feature, for the function

provides reliable method to discover the network nodes and their variation. In this

case, the nodes of an RPR can be automatically added, deleted and switched.

There may be more than one EMR0 in equipment, so an NE may have more than one

RPR node. The plug and play feature allows adding or deleting nodes without

affecting the services on an RPR.

To increase or decrease the total bandwidth of an RPR, use the LCAS function. The

LCAS features adding and reducing bandwidth dynamically without affecting existing

services.

4. Spatial Reuse

The stripping of unicast frames at the destination station realizes spatial reuse on an

RPR. As shown in Figure 6-10, the bandwidth of a ring is 1.25 Gbit/s. Traffic 1

transferred from node 1 to node 4 is stripped from the ring at the destination node 4.

After the arrival of traffic 1 at node 4, traffic 2 can be transferred from node 4 to node 3,

by occupying the link capacity that would have been occupied by traffic 1 if it is not

stripped at node 4.

6-12

Traffic 1

1.25 Gbit/s

Node 2

Dual-ring

Node 4

2.5 Gbit/s RPR

Traffic 2

1.25 Gbit/s

Bandwidth of single ring is

1.25Gbit/s

Node 3

Figure 6-10 Spatial reuse

5. Fairness Algorithm

The outer ring and the inner ring of an RPR support independent weighted fairness

algorithm. The fairness algorithm assures access of the low-class B_EIR and C

services. The weight of the fairness algorithm is provisionable to decide the access

rate of a node. A node needs to set weights at the outer and the inner rings, and the

two weights decide the bandwidth of low-class services upon bandwidth contention.

As shown in Figure 6-11, the outer ring weights of nodes 2, 3 and 4 are 1. Suppose the

available bandwidth on the outer ring for low-class services is 1.2 Gbit/s, the fairness

algorithm will allocate 400 Mbit/s for the low-class services from nodes 2, 3 and 4 to

node 1 respectively. Figure 6-12 shows a fairness algorithm with different weights: the

weights of nodes 2, 3 and 4 on the outer ring are 1, 3 and 2 respectively. The fairness

algorithm allocates 200 Mbit/s for node 2, 600 Mbit/s for node 3 and 400 Mbit/s for

node 4.

6-13

Weight

Node2

Node3

Node 2

Node4

Node 3

Node 1

Dual-ring

2.5 Gbit/s RPR

Node 4

Node 6

Traffic

Bandwidth

400 Mbit/s

Node 5

400 Mbit/s

Figure 6-11 Fairness algorithm when the weight is 1

Node

Weight

Node2

Node3

Node 2

Node4

Node 3

Node 1

Dual-ring

2.5 Gbit/s RPR

Node 4

Node 6

Traffic

Bandwidth

200 Mbit/s

Node 5

600 Mbit/s

400 Mbit/s

Figure 6-12 Fairness algorithm when the weights are different

6.2.2 Application

The EMR0 and EGR2 boardS support the application of EVPL and EVPLAN services.

6-14

The EVPL service supports traffic classification based on port or port + VLAN, and

encapsulates and forwards the traffic in the form of MPLS MartiniOE. Figure 6-13

illustrates the accessing, forwarding and stripping of a unidirectional EVPL service.

Node 2 inserts Tunnel and VC labels to the packet, sends it to the RPR. Node 3

forwards the packet and the destination node 4 strips it. Figure 6-14 illustrates the

EVPL service convergence, implementing traffic classification through port + VLAN,

so that services can be concentrated at the GE port of node 1.

Node 1

Dual-ring

2.5 Gbit/s RPR

FE/GE

Node 2

Node 4

Action

Stripping

LSP

Tunnel

100

Node 3

Action

Insertion

100

Tunnel

100

Destination

Node 4

Action

Forwarding

Figure 6-13 EVPL accessing, forwarding and stripping

6-15

Content .. 3 4 5 6 ..