Illustrated TCP/IP by Matthew G. Naugle Wiley Computer Publishing, John Wiley & Sons, Inc. ISBN: 0471196568   Pub Date: 11/01/98

Chapter 250
Unicast (versus Multicast)

For example, let’s say that the end of the month sales report is complete and needs to be transmitted to 200 file servers around the country. With unicast addressing, one could write a simple script that would initiate a file transfer to each of the 200 file servers. If the file is 2 MB in length, you now have a 400 MB file transfer that will unnecessarily consume both time and bandwidth.

Enabling multicast would push that one file to the 200 file servers simultaneously, reducing the number of file transfers from 200 to 1, and the size of the file from 400 MB to 2 MB.

One example in which multicast file transfer is used extensively is software distribution. Microsoft and other application companies upgrade their software at least twice a year—a daunting task in large environments. If the upgrade is 20 MB and must be distributed to 50,000 desktops, it would require 1000 billion bytes to be delivered. Assuming the transfer occurs at 512,000 bits per second, the upgrade would take 180 days to complete. And just when you completed one upgrade, the next one is ready.

However, using reliable multicasting, the file could be delivered to 50,000 desktops as a single stream. This means one transfer looks like 50,000. With multicast, the same transmission would complete in 5.2 hours. (This number assumes no real-time errors such as retransmissions. Retransmission costs are variable depending on the number of clients requesting, the status of the lines, etc.) Even so, the worst cost is less than 25 percent of the original pass. The example is theory, but if put into practice I believe that it would be around the number indicated. Even if it took 10 hours to complete, look at the alternative.

Unicast (Versus Multicast)

Therefore, the advantages to multicast-enabled applications and networks are time savings and scaleable bandwidth.