Exploring multicast IP in Linux
Listing 3 shows the steps for enabling multicast transmissions. As you can see, multicasting is enabled for individual interfaces and for the unicast interface.
Enabling Multicast Transmission
01 # set plumbing mfea4 disable false 02 # set plumbing mfea4 interface eth0 vif eth0 disable fasle 03 # set plumbing mfea4 interface eth1 vif eth1 disable fasle 04 # set plumbing mfea4 interface register_vif vif register_vif disable fasle 05 # commit
The final step in the configuration is to enable the PIM-SM protocol. Start by loading the daemon of the PIM protocol:
# set protocols pimsm4 disable false # commit
The #commit command will be executed with a slight delay because of the initiation of the process responsible for the service of the multicast routing protocol. Subsequent commands, presented in Listing 4, configure the PIM-SM protocol.
The commands in Listing 4 activate the PIM-SM protocol service on individual interfaces, as well as on the virtual interface register_vif, which is used to transfer data over a unicast tunnel from the source to the rendezvous point. Additionally, the address of the rendezvous point, which is 192.168.3.1 in this case, is allocated. The group-prefix setting denotes the range of multicast addresses serviced by a given rendezvous point.
01 # set protocols pimsm4 interface eth0 vif eth0 disable false 02 # set protocols pimsm4 interface eth1 vif eth1 disable false 03 # set protocols pimsm4 interface register_vif vif register_vif disable false 04 # set protocols pimsm4 static-rps rp 192.168.3.1 group-prefix 22.214.171.124/4 05 # commit
The configuration presented so far makes it possible to transfer data from the source to particular receivers. The receivers, however, must be able to inform routers that they are interested in receiving multicast transmissions. As we described earlier, this information passes through the network via IGMP, so you must configure IGMP on the routers that have local receivers.
At this point, you might be wondering why it is necessary to configure IGMP through XORP when the kernel already supports IGMP.
The problem is that the implementation of IGMP that is included in the kernel does not provide the server side of IGMP, which means that the built-in kernel implementation will not forward information in IGMP messages to the multicast routing protocol.
As in the case of PIM-SM or OSPF, the IGMP configuration requires activation of a daemon responsible for the service of the protocol:
# set protocols igmp disable false # commit
In addition, you must indicate the interfaces serviced by IGMP:
# set protocols igmp interface eth2 vif eth2 disable false # commit
Putting It Together
Figure 4 shows the whole transaction at a glance. As you can see, the computer on the left starts by sending a video sequence to the address 126.96.36.199. Router A, directly connected to the source, starts sending data to the rendezvous point through a unicast tunnel. The application on the receiving end generates an IGMP Report message. This message is processed by the router, which is followed by a Join message of the PIM-SM protocol sent toward Router C, which is acting as the rendezvous point RP. On receiving the message, Router C sends the multicast transmission in the direction of this receiver, plus any other receivers in the group that will receive the multicast data.
Buy this article as PDF
Powerful man-in-the-middle attack is now targeting online shopping.
Another high-profile coder says the kernel team needs a kinder, gentler culture.
Bug database has a bug of its own that could allow an intruder to create an unauthorized account.
Report focuses federal resources on achieving universal Internet access.
Leading browser makers say “no” to porous encryption algorithm
Report from the X-Force group says attackers are using TOR to hide their crimes
Future Firefox extensions will be compatible with Chrome.
Better read this if you bought your computer before 2011
Users should upgrade to the new version as soon as possible
Xen project announces a privilege escalation problem for Qemu host systems