Muticast design issue

[sateeshk10]

Hi,

I want enable multicast in my network ..Below is my requirements

1) I am planning to enable multicast routing in my core switch

2) ACCess switch1 connted to core SW trgh trunk

3)My multicast server has connected to access swtich. (Vlan 10)

What config is required in access swtich to pass multicast traffic?

If access sw2 wants to comunicate trgh vlan 20( configured in core switch only)

Access sw2 also connected to core switch as trunk.

pls let me know the config...which i need to be done in acess swtcihes and core swtich.

Regards

sateesh

[francisco_1]

sateesh,

the configuration depends on what you are trying to do. are you going to route the multicast traffic between the vlans on the core or just have the multicast traffic within a single vlan? On the access layer you normally will need to enable igmp snooping on the access layer to relay the multicast traffic to your core switch when a client join the multicast group. on the core switch if the multicast traffic is staying within a vlan, then no need to enable multicast routing. Just make sure igmp snooping is enable on the vlan SVI.

see this http://www.cisco.com/en/US/tech/tk828/technologies_tech_note09186a0080094821.shtml

Francisco.

[sateeshk10]

Hi,

Thanks for your crystal clear reply..its pretty good..

Multicast clients will be on difrrent VLANS..is it suggestable??

2960 is suggestable in datacenter as access layer???

I am planning to put core sw (6500)??

I need Multicast config in access swtiches and coreswtiches??

Old Setup:

Core swtiches 6509 ad access switch 3750 runing L3 on access swtiches..Multicast routing on core and access also..Now this i want to make my access as L2...So i need above info...

Regards

sateesh

[francisco_1]

sateesh, i cannot recommend what switch to use on your access layer but 2960's are good switch to start with. You can either enable multicast on the access layer if you have the hardware to do so. if not then no need.

To start with basic multicast to get you going.

1, enable igmp snooping for your active vlans on the 2960.

2 enable on core switch ip mutlicast routing.

3 enable ip pim sparse-mode under the vlans (vlan multicast will be acvtive on) SVI on the core switch.

4, If you having multiple core switches, then enable on of them as RP.

for more info, visit the cisco website/

Francisco

[sateeshk10]

Hi,

Thanks for info..

enable igmp snooping for your active vlans on the 2960??? Where r to enable???

4, If you having multiple core switches, then enable on of them as RP.??

I have two core switches..ONe is HSRP pri and other one is secondary..NOw pls let me know where to enable..

Regards

sateesh

[francisco_1]

By default, IGMP snooping is globally enabled on the 2960 switch so no need .

Enable RP on either on of your core switch maybe preferabely the active hsrp. The command is "ip pim rp-address [ip address of maybe a loopback interface or SVI on the prefered switch]" Also enter this command on the other core switch as well.

Francisco

[sateeshk10]

Hi..

Do we need "ip pim rp-address..As we can use ip pim spare-dense mode...

If we are only spare then we need rp point..if it dense mode means it willl flodd the multicast traffic.. right

Offcouse i ma not sure..I need some clafication on abv command

Regards

sateesh

[francisco_1]

the RP works with sparse-mode only so the multicast traffic (source traffic) is sent towards the RP (RP will forwards traffic to destination ONLY interested in joining the multicast group) and not flooded to all devices unnessary. with dense mode you flooding and prune the traffic to all devices even devices not interested in receiving the multicast traffic. with sparse-mode you have more control. with sparse-dense-mode, you are saying if there is no RP, then send the traffic as dense mode. if RP exist then send as sparse-mode.

multicast is a very deep subject and we can go on and on.

i strongly suggest you start reading the basics. start here http://www.cisco.com/en/US/tech/tk828/tsd_technology_support_protocol_home.html

Francisco

[abdel_n]

Hi Sateesh,

Let's try to make difference between multicast over LAN network (Layer2 as well as Layer3 capable switches and multicast servers and clients) and multicast over Layer3 network generally in core network routers.

Here is a couple of brief definitions necessary to understand what, when and where.

1-For your LAN the goal is to make sure multicat servers and clients can advertise and receive respectively from the LAN gateway or gateways using IGMPv1/v2/v3 or CGMP

- Layer2 switches doesn't understand the "language" of multicast (IGMP) therefore cannot switch properly (will broadcast to all ports), LAN gateway will translate IGMP to them using CGMP (so only clients who need multicast will receive it).

- Layer3 capable switches understand IGMP, so will listen to IGMP traffic (without participating) and switch accordingly.

- RGMP is similar to IGMP but optimize switching to router ports.

2-In the core network multicast need to be routed using PIM:

- PIM Dense mode assumes that everyone in the network wants to receive multicast, so flood it everywhere and it is up to other routers to prune it if no requests.

- In the contrary, PIM Sparse mode will not send anything until it was requested explicitly. Hosts, whatever multicast requested, form shared trees with RP (Rendez-vous Point) which in turns play the proxy role and form source routed trees with the source (router connected to source), this is done the time to request multicast after that they can communicate directly (router connected to source and router connected to clients).

In your case, your core switch (route between VLANs) can be the gateway for your LAN where PIM is enabled and SW1/SW2 (c2960) your access switches where IGMP snooping is enabled.

I hope this will help you in your design.

Abdel

[sateeshk10]

Hi abdel,

Thanks for your brief info..

I am going to enable following commands on core switches and access switches

Core:

ip multicast routing

int vlan 10

ip pim spare-dense

Acesss:

igmp snooping

Do i need to enable the ip pim sparse-dense in uplinks also???

Regards

sateesh kumar.k

[abdel_n]

Hi Sateesh,

-*- On access switches (No PIM routing, but multicast swiching):

Acoording to the C2960 configuration guide: http://www.cisco.com/en/US/docs/switches/lan/catalyst2960/software/release/12.2_46_se/configuration/guide/swigmp.html#wp1089644

IGMP snooping is enabled on all VLAN interfaces by default

(config)#igmp snooping

-*- The following are needed steps to configure PIM Sparse-Dense mode with static RP (on routing devices):

1) Enable sparse-dense mode globally:

(config)#ip multicast-routing

2) enable sparse-dense on PIM routers:

The interface command "ip pim sparse-dense" have to be enabled on all routing interfaces through which the multicast traffic can possibly be processed, in the path from sources to destinations, so if your core switch routes between VLANs 10 and 20, you should enable PIM sparse-dense on each of the SVI (Switch Virtual Interface).

int vlan 10

ip pim spare-dense

int vlan 20

ip pim spare-dense

3) RP (Rendez-vous Point) assignment:

Configure statically all core devices participating in multicast routing with RP ip address (RP itself included)

(config)#ip pim rp-address X.X.X.X

with XX.X.X an advertised loopback interface on the RP.

By default the RP will be responsible for all multicast groups.

Abdel

[sateeshk10]

Hi,

I hope third point is not required in my scenario right as i am going for both modes spare-dense mode ?

Regards

sateesh

[abdel_n]

In PIM-Sparse mode (not dense mode) every router MUST learn statically or dynamically the IP address of RP, it is not an option. PIM-sparse-dense mode is mainly used for dynamic RP because it is more flexible and allow an interface to operate on a per-group basis in either sparse or dense mode.

When using static RP, you can just use PIM-sparse mode because RP IP is manually configured, simple and straightforward for small networks, however in relatively big networks static RP is a burdensome procedure and prone to errors.

With dynamic RP, PIM routers must join 224.0.1.40 to be able to learn RP dynamically, BUT to join 224.0.1.40 they have to know the RP ???

=> Here comes PIM sparse-dense mode in which RP will flood 224.0.1.40 (as well as group 224.0.1.39) using dense mode, so PIM routers join those groups an can learn dynamically RP address by listening to 224.0.1.40 and further use PIM-Sparse mode for remaining multicast groups.

[sateeshk10]

Hi,

Small silly question where can i get RP IP( It is not but where my multicast applications are runing) Right. That IP i have to give as RP IP..

Regards

sateesh

[abdel_n]

RP doesn't have to do with the multicast application, it is a concept related to PIM-Sparse mode (sparse-dense as well) and how the multicast traffic is routed between PIM routers, the core in your case.

In other words RP will be one of your core devices where routing between VLANs is performed, Cat6500:

ip multicast-routing

int vlan 10

ip pim spare-dense

int vlan 20

ip pim spare-dense

int loopback0

ip address

ip pim rp-address