Day2_Multicast

7/31/2019 Day2_Multicast

1/24

FCE MulticastAdam Allen

IN CONFIDENCE

Jan 2009


2/24

FCE Multicast

CORE DESIGN


3/24

British Telecommunicationsplc

Multicast vs Unicast

Multicast has been a significant development on BT MPLS. There aretwo flavours of Multicast: Intranet between sites of a common customer Extranet Provider/Member model

Multicast Traffic is not label switched across the core network. However VPNs are still used to control reach ability of Multicast

Sources and RPs

Access to the Multicast service needs to be provisioned on the PErouter giving strong Service Level Entitlement.

Start Simple 1 Multicast Provider in the network providing 1 Multicast Service with 1 Multicast Group


4/24


Default - Multicast Distribution Tree

CE

CE1-LFA

P

RR

PE

CE1-NYK

PE

P

P

P

CE1-HKK

239.232.56.1

The provider is allocated a Default MDT address

(239.232.56.1), configured against the London PE

As clients subscribe to the multicast service

the Default MDT is extended to the client

serving PE routers.

The MDT addresses are controlled by

BT and are not visible or connected to

the customer. The default MDT is

always active.

CE2-NYK


5/24


Transport VRFs

The mechanism by which Default MDTs are extended to

PE routers are by the configuration of Transport VRFs

The Transport VRF is configured on an egress client PE

router and is associated with the Default MDT

The Tranposrt VRF is shared by all customers on that PE

router who subscribe to the Multicast service.

ip vrf fe-ctv-adammcast-1

rd 12641:3824211

mdt default 239.232.56.1

mdt data 232.232.156.4 0.0.0.0

threshold 1


6/24


Data - Multicast Distribution Tree

CE

CE1-LFA

P

RR

PE

CE1-NYK

PE

P

P

P

CE1-HKK

239.232.56.1

Initially Multicast traffic is sent via the Default MDT until the traffic rate crosses the

threshold (1Kbits).

Then a Data MDT (239.2323.156.4) will

dynamically created to carry traffic to

the interested PEs. Traffic remains on

the Default MDT for 13 seconds afterthe Data MDT is created.

CE2-NYK

239.232.156.4


7/24


C packets / P packets

CE

P

RR

PE

PE

P

P

P

239.232.56.1

239.232.156.4

Source

155.0.0.2Destination

224.31.1.1

GRE Packet

Source

212.31.220.44

Destination

239.232.156.4

Source


224.31.1.1

GRE Packet

Source

212.31.220.44

Destination

239.232.156.4

Source


224.31.1.1

Source

155.0.0.2

Destination

224.31.1.1


8/24


Extranet Functionality

CE

CE1-LFA

P

RR

PE

CE1-NYK

PE

P

P

P

CE1-HKK

239.232.56.1

CE2-NYK

Multicast Extranet is a

recent feature development

on the PE router to allow a

single Transport VRF to be

shared by multiple

customers on the same PE

router.

Service 3

Service 3

Service 3

239.232.156.4

Transport VRF

ip vrf fe-ctv-adammcast-1

rd 12641:3824211

mdt default 239.232.56.1

mdt data 232.232.156.4 0.0.0.0 threshold 1

Client 1s VRF

ip vrf ex-finance-126414rd 12641:4

Client 2s VRF

ip vrf ex-finance-126415rd 12641:5


9/24


Extranet

ip multicast vrf ex-finance-126414

rpf select vrf fe-ctv-adammcast-mtv1 group-

list ACL-FE-mcast-adammcast_1

The rpf select command creates a link between the client

VRF and the Transport VRF when accessing the multicast

groups defined in the ACL

ACL-FE-mcast-adammcast_1

224.31.1.1

Only customers subscribing to this multicast service will

have the rpf select command configured.


10/24


Transport VPN Mapping.

Transport VPN Mapping is a trade-off between core-state and core-bandwidth.

If the Multicast-group 232.31.1.1 supports a high bandwidth feed thiscould be carried multiple times across the core network.

The solution for this situation is to create 3 Transport VPNs for thisprovider.

The previous examples show Service 3

Service 1

232.31.1.1

232.31.1.2

Service 2

232.31.1.1232.31.1.4

Service 3

232.31.1.1

Transport VPN 1groups 232.31.1.1

default mdt 239.232.56.1

data mdt 239.232.156.4



data mdt 239.232.156.5



data mdt 239.232.156.6


11/24


Transport VPNs

CE

CE1-LFA

P

RR

PE

CE1-NYK

PE

P

P

P

CE1-HKK

239.232.56.1

CE2-NYK

In this example CE1 has subscribed to

Service 3.

In this example CE1 has subscribed to

Service 3, and CE2 has subscribed to

Service 2.

239.232.156.4

239.232.56.2

239.232.156.5

Traffic associated with

232.31.1.4 will never head

to Hong Kong as no clients

subscribe to Service 2

Traffic associated with

232.31.1.1 is only

transported to PE1-NYK

once.

Service 3

Service 2

Service 3


12/24


RP

CE

CE1-LFA

P

RR

PE

CE1-NYK

PE

P

P

P

CE1-HKK

239.232.56.1

Head-end CEs router are configured to

act as a Rendevouz Point. The RP is

configured as an Anycast RP

239.232.156.4

Service 3

RP 155.155.155.1


13/24


RP

ip pim vrf ex-finance-126414rp-address 155.155.155.1 ACL-FE-mCast-Groups

The Anycast RP is configured for each VRF on the client PE router.

ACL-FE-mCast-Groups matches the aggregated set of FCE group

addresses.

ip pim vrf fe-ctv-adammcast-mtv1rp-address 155.155.155.1 ACL-FE-mCast_adammcast

The Transport VRF is configured to maps its own Multicast groups tothe Anycast RP

Routing to the RP and Source is provided by importing a per-service-RP-RT and per-service-Source-RT


14/24


Live-Live

CE

CE1-LFA

P

RR

PE

CE1-NYK

PE

P

P

P

239.232.56.6

CE2-NUT

239.232.156.7

239.232.56.8

239.232.156.9

Feed-A and Feed-B will map

to different Transport VPNs.

Feed-A and Feed-B will mapto different RP addresses.

Service 4

Service 4


15/24

FCE Multicast

CLIENT DESIGN


16/24



17/24


VLAN 180

Service/VLAN

mapping

Multicast Services are associated with specific VLANs

Unlike Unicast this extends to the configuration (SLE ACL, IGMPStatic-group statements)

Services are only available via the VLANs they have been orderedby

Services may be delivered via the Generic FE VLAN (170) or viaseparate VLANs

VLAN 170


18/24


BT Managed Switch VLANS

In addition to the list of VLANs presented earlier as part of

the FCE High Level Design additional VLANs are

available.

The following VLANs do not have default port allocations

All are allocated over the ISL.

SPARE VLANs

VLANs 220 - 235

VLANs 237 - 249

ThomsonOne VLAN 210

Instinet VLAN 110

State Street VLAN 120

Comstock VLAN 236


19/24


Always On

By default all Multicast Groups the customer has subscribed to willbe explicitly configured on the client CE router

However it is possible for the customers to order the service withoutstatic IGMP joins. In this case the customer will either send IGMP joins or speak PIM in

order to control the Multicast service. DR Priority can be modified on the MPLS CEs

Default 200 Primary / 100 Secondary

VLAN 170

Interface GigabitEthernet0/0.170

ip igmp static-group 224.31.1.1


20/24


SLE

The CE router has a set of ACLs configured which provide amapping between the groups to RP

The groups listed in the ACL are the groups which the customer hassubscribed to.

PIM/IGMP Joins received for groups not listed in the ACL will bemarked invalid.

Service Level Entitlements are also enforced at the PE router.

VLAN 170


21/24


VLAN 180

PIM

Upto 10 PIM Talkers can be specified on the customer LAN. The PIM

talkers will be configured into an PIM Boundary ACL. PIM neighbourships are established over the ICRL link if provided.

Ordering PIM requires the use of Customer LAN Routing.

RIP is selected by default

The RP address is redistributed into the Customer LAN Routing

VLAN 170


22/24


Live Live

Live-Live is carried out by using separate S,G pairs for the A-Feed

(Red) and the B-Feed (Blue) Live-Live services use a different RP to Live-Standby services. Where a customer is not a pim talker/using static joins the intial joins

for the B-Feed will be directed to the Designated Router The PIM join redirects to the secondary router The Traffic flow does not converge

VLAN 170

A B


23/24


SSM

Source Specific Multicast is supported with the BT MPLS

network designs.

Where the service is not configured as an Always On

service.

IGMP Version 2 does not allow the source address to be

specified.

IGMP Version 3 or PIM needs to be used


24/24


UDP Broadcast/Multicast Conversion

Broadcast Multicast conversion allows for providers to

send UDP broadcast traffic at the head-end.

The Head-end CE router converts the traffic to a Multicast

Feed before sending into the BT MPLS network.

At the Member CE router the traffic is re-converted back to

UDP traffic before broadcasting on the LAN

Day2_Multicast

Documents