Why Do You Charge Multicasting? 535
Problems with Unicast and Advertisement Methods
Why not use unicast or advertisement methods to accelerate a bulletin from one antecedent to abounding destinations?
Figure 16-1 shows a video server as a antecedent of a video appliance and the video abstracts that needs
to be delivered to a accumulation of receivers—H2, H3, and H4—two hops abroad beyond a WAN link.
Figure 16-1 Unicast
The unicast adjustment requires that the video appliance accelerate one archetype of anniversary packet to every
group member’s unicast address. To abutment full-motion, full-screen viewing, the video stream
requires about 1.5 Mbps of bandwidth for anniversary receiver. If alone a few receivers exist, as
shown in Figure 16-1, this adjustment works accomplished but still requires n × 1.5 Mbps of bandwidth, where
n is the cardinal of accepting hosts.
Figure 16-2 shows that as the cardinal of receivers grows into the hundreds or thousands, the load
on the server to actualize and accelerate copies of the aforementioned abstracts additionally increases, and replicated unicast
transmissions absorb a lot of bandwidth aural the network. For 100 users, as adumbrated in the
upper-left bend of Figure 16-2, the bandwidth adapted to accelerate the unicast manual increases
to 150 Mbps. For 1000 users, the bandwidth adapted would access to 1.5 Gbps.
H1 H2
Frame
Relay Cloud
1.5 Mbps 3 4.5 Mbps
Video
Server
1.5 Mbps 1 1.5 Mbps
1.5 Mbps 2 3 Mbps
1.5 Mbps 1 1.5 Mbps
H3 H4
1.5 Mbps 1 1.5 Mbps
1.5 Mbps 1 1.5 Mbps
536 Affiliate 16: Introduction to IP Multicasting
Figure 16-2 Unicast Does Not Scale to Large Numbers of Receivers
You can see from Figure 16-2 that the unicast adjustment is not scalable. Figure 16-3 shows that the
broadcast adjustment requires manual of abstracts alone once, but it has some austere issues. First, as
shown in Figure 16-3, if the receivers are in a altered advertisement area from the sender, routers
need to advanced broadcasts. However, forwarding broadcasts ability be the affliction accessible solution,
because broadcasting a packet to all hosts in a arrangement can decay bandwidth and increase
processing amount on all the arrangement accessories if alone a baby accumulation of hosts in the arrangement actually
needs to accept the packet.
Figure 16-3 Advertisement Wastes Bandwidth and Increases Processing Amount on CPU
H1 H50
ATM
Cloud
1.5 Mbps 100 150 Mbps
Video
Server
1.5 Mbps 50 75 Mbps
1.5 Mbps 50 75 Mbps
1.5 Mbps 1 1.5 Mbps
1.5 Mbps 1 1.5 Mbps
H51 H100
1.5 Mbps 1 1.5 Mbps
1.5 Mbps 1 1.5 Mbps
. . . . . . . . . . . .
Frame
Relay Cloud
?
1.5 Mbps
Video
Server
I didn’t ask for this
traffic. Why should I
process it?
H1 H2
?
1.5 Mbps ?
1.5 Mbps
1.5 Mbps 1.5 Mbps
H3 H4
1.5 Mbps 1.5 Mbps
Why Do You Charge Multicasting? 537
How Multicasting Provides a Scalable and Manageable Solution
The six basal requirements for acknowledging multicast beyond a baffled arrangement are as follows:
■ A appointed ambit of Layer 3 addresses that can alone be acclimated by multicast applications must
exist. A arrangement ambassador needs to install a multicast appliance on a multicast server
using a Layer 3 multicast abode from the appointed range.
■ A multicast abode charge be acclimated alone as a destination IP abode and accurately not as a
source IP address. Unlike a unicast IP packet, a destination IP abode in a multicast packet
does not specify a recipient’s abode but rather signifies that the packet is accustomed multicast
traffic for a specific multicast application.
■ The multicast appliance charge be installed on all the hosts in the arrangement that charge to receive
the multicast cartage for the application. The appliance charge be installed appliance the same
Layer 3 multicast abode that was acclimated on the multicast server. This is referred to as
launching an appliance or abutting a group.
■ All hosts that are affiliated to a LAN charge use a accepted adjustment to account a Layer 2
multicast abode from the Layer 3 multicast abode and accredit it to their arrangement interface
cards (NICs). For example, if assorted routers are affiliated to an Ethernet articulation and all
of them are appliance the OSPF acquisition protocol, all the routers on their Ethernet interfaces will
also be alert to the Layer 2 multicast abode 0x0100.5e00.0005 in accession to their
Burned-In Addresses (BIA). This Layer 2 multicast abode 0x0100.5e00.0005 is calculated
from the multicast Layer 3 abode 224.0.0.5, which is aloof for the OSPF routing
protocol.
■ There charge be a apparatus by which a host can dynamically announce to the affiliated router
whether it would like to accept the cartage for the installed multicast application. The Internet
Group Management Agreement (IGMP) provides advice amid hosts and a router
connected to the aforementioned subnet. The Cisco Accumulation Management Agreement (CGMP) or IGMP
snooping helps switches apprentice which hosts accept requested to accept the cartage for a specific
multicast appliance and to which about-face ports these hosts are connected.
■ There charge be a multicast acquisition agreement that allows routers to advanced multicast traffic
from multicast servers to hosts after arduous arrangement resources. Some of the multicast
routing protocols are Distance Vector Multicast Acquisition Agreement (DVMRP), Multicast Open
Shortest Path Aboriginal (MOSPF), and Agreement Independent Multicast close approach (PIM-DM)
and dispersed approach (PIM-SM).
This affiliate discusses the aboriginal bristles bulleted items, and Affiliate 17, “IP Multicast Routing,” covers
the multicast acquisition protocols.
538 Affiliate 16: Introduction to IP Multicasting
Figure 16-4 shows how multicast cartage is forwarded in a Layer 3 network. The purpose of this
illustration is to accord you an overview of how multicast cartage is forwarded and accustomed by
selected hosts.
Figure 16-4 How Multicast Delivers Cartage to Selected Users
Assume that a video multicast appliance was installed on the video server appliance the special
Layer 3 multicast abode 225.5.5.5. Hosts 1 to 49, amid beyond a WAN link, are not interested
at this time in accepting cartage for this application. Hosts 50 to 100 are absorbed in receiving
traffic for this appliance and barrage this appliance on their PCs. When the host launches the
application, the host joins the group, which agency that the host now wants to accept multicast
packets beatific to 225.5.5.5. Hosts 50 to 100 accompany accumulation 225.5.5.5 and announce to R2 their admiration to
receive cartage for this multicast appliance by appliance IGMP. The multicast appliance calculates
the Layer 2 multicast abode 0x0100.5e05.0505 from the Layer 3 multicast abode 225.5.5.5,
and NICs of hosts 50 to 100 are alert to this abode in accession to their BIAs.
A multicast acquisition agreement is configured on R1 and R2 so that they can advanced the multicast
traffic. R2 has one WAN articulation affiliated to the Frame Relay billow and two Ethernet links
connected to two switches, SW2 and SW3. R2 knows that it has on both Ethernet links hosts that
would like to accept multicast cartage for the accumulation 225.5.5.5 because these hosts accept indicated
their admiration to accept cartage for the accumulation appliance IGMP. Both switches accept additionally abstruse on which
Frame
Relay Cloud
1.5 Mbps
209.165.201.1/27
Video
Server
SW1
SW2
R1
R2
1.5 Mbps
1.5 Mbps
1.5 Mbps 1.5 Mbps 1.5 Mbps
MM: 0x0100.5e05.0505 MM: 0x0100.5e05.0505 MM: 0x0100.5e05.0505
H1 H50 H51 H100
Multicast Packet
Destination Address: 225.5.5.5
Source Address: 209.165.201.1
SW3
. . . . . . . . . . . .
MM Multicast MAC Address
Why Do You Charge Multicasting? 539
ports they accept hosts that would like to accept the multicast cartage for this appliance by using
either CGMP or IGMP snooping.
A multicast packet campaign from the video server over the Ethernet articulation to R1, and R1 assiduously a
single archetype of the multicast packet over the WAN articulation to R2. When R2 receives a multicast packet
on the WAN articulation with the destination abode 225.5.5.5, it makes a archetype of the packet and
forwards a archetype on anniversary Ethernet link. Because it is a multicast packet for the accumulation (application)
225.5.5.5, R2 calculates the Layer 2 destination multicast abode of 0x0100.5e05.0505 and uses
it as the destination MAC abode on anniversary packet it assiduously to both switches. When the switches
receive these packets, they advanced them on adapted ports to hosts. When the hosts accept the
packets, their NICs analyze the destination MAC abode with the multicast MAC abode they
are alert to, and, because they match, acquaint the college layers to action the packet.
You can see from Figure 16-4 that the multicast cartage is beatific already over the WAN links and is
received by the hosts that accept requested it. Should added hosts appeal to accept the same
multicast traffic, neither the multicast server nor the arrangement assets would acquire any additional
burden, as apparent in Figure 16-5.
Figure 16-5 Multicasting Is Scalable
Assume that hosts 1 to 49 accept additionally adumbrated their admiration to accept cartage for the multicast group
225.5.5.5 appliance IGMP. R2 is already forwarding the cartage to both switches. Either CGMP or
Frame
Relay Cloud
1.5 Mbps
209.165.201.1/27
Video
Server
SW1
SW2
R1
R2
1.5 Mbps
1.5 Mbps
1.5 Mbps 1.5 Mbps 1.5 Mbps 1.5 Mbps
MM: 0x0100.5e05.0505 MM: 0x0100.5e05.0505 MM: 0x0100.5e05.0505 MM: 0x0100.5e05.0505
H1 H50 H51 H100
SW3
I aloof joined!
Send me traffic
for 225.5.5.5.
Multicast Packet
Destination Address: 225.5.5.5
Source Address: 209.165.201.1
. . . . . . . . . . . .
MM Multicast MAC Address
IGMP concern can advice SW2 (shown in Figure 16-5) apprentice that hosts 1 to 49 accept additionally requested
the multicast cartage for the accumulation so that it can alpha forwarding the multicast cartage on ports
connected to hosts 1 to 49. The added 49 users are now accepting multicast traffic, and the load
on the multicast server, amount on added arrangement devices, and appeal for bandwidth on the WAN
links abide the same. The amount on SW2 apparent in Figure 16-5 increases because it has to make
49 added copies of the multicast cartage and advanced it on 49 added ports; however, it is now
operating at the aforementioned akin as the added switch. You can see that IP multicast is scalable.
Although multicast offers abounding advantages, it additionally has some disadvantages. Multicast is UDPbased
and appropriately unreliable. Lack of TCP windowing and “slow start” mechanisms can aftereffect in
network congestion. Some multicast agreement mechanisms occasionally accomplish duplicate
packets and bear packets out of order.