Unicast

The network tech­nol­o­gy knows different methods for receiving a message from the addressee to the recipient(s). While a broadcast allows you to direct your data to all users in a network, unicast helps with a targeted address. Multicast works in a similar way, but with the message going to several specific re­cip­i­ents at the same time. What exactly is a unicast and how do unicast and multicast differ from one another?

No matter which form of com­mu­ni­ca­tion you’re using, if you wish to reach a specific contact, you have to know how to address them. Name, telephone number, postal address – they all help us to direct our in­for­ma­tion to the correct re­cip­i­ents. This is no different with network tech­nol­o­gy. The IP and MAC address help identify the re­cip­i­ents here.

In the OSI model, the unicast is on the network layer (layer 3) and is therefore a routing diagram. A header with the address data is added to the in­for­ma­tion that is to be sent, re­gard­less of whether it is an e-mail, a file or a simple website request. For simple pro­ce­dures, the address of the recipient in the network is located in the header. The packet will be sent in this direction.

However, com­mu­ni­ca­tion does not always take place within a closed network only where the router can reach each end point directly. Unicasts can also be sent between different (sub-)nets. IP routing tech­nol­o­gy is relied upon for this. This ensures that every hub knows which path the data packet must take for it to reach the le­git­i­mate recipient. Routers (the network hubs) use routing tables for this. To create these, the OSPF and RIPv2 protocols are used in par­tic­u­lar.

In contrast to the IPv4 addresses that have been used up to now, the new IPv6 has earmarked a special feature for unicast: specific address ranges are reserved for the various types of com­mu­ni­ca­tion. Multicast addresses are located in a different address range to the unicast addresses. There is also dif­fer­en­ti­a­tion between various types of unicast addresses, which are rep­re­sent­ed by a prefix at the start of the address.

Link local des­ig­nates local, self-contained networks. No redi­rec­tion to other networks by a router is required here; the range fe80::/10 is reserved for this. The first 10 bits of the address are intended for the prefix. 64 bits follow, which are all set to 0. The address ends with a 54-bit long range, which discloses the interface ID. This iden­ti­fies the client uniquely in the local network.

In contrast to the link local addresses, a unique local unicast can be redi­rect­ed through a router. Nev­er­the­less, these unique addresses are always assigned in­ter­nal­ly (in a set network range, e.g. the network of a large company). The Internet En­gi­neer­ing Task Force (IETF) has provided the range fc00::/7 for this. However, only the part fd00::/8 has been provided specif­i­cal­ly for the unique local unicast; no decision has yet been made about the range fc00:/8. A 40-bit part follows the prefix, which contains a randomly generated site ID. The end forms a 16-bit long subnet-ID.

With the global unicast, it is also possible to send messages worldwide over unicast with IPv6 . The unique addresses are allocated globally, making it possible to reach someone in a very targeted way. This is virtually the standard case for an IPv6 address. The first part of the address is des­ig­nat­ed as a location prefix or public topology; it depends on the internet provider. In­for­ma­tion then follows on the subnet and the actual client. As with sub­net­ting, the last part of the address (interface ID) may only be assigned once within the subnet. This is the only way it can be ensured that a unicast also actually reaches the correct target.

Both types of com­mu­ni­ca­tion, unicast and multicast, have clear sim­i­lar­i­ties, par­tic­u­lar­ly when you compare them with broadcast. The broadcast has its own address. All users on the network recognize it and can therefore respond to the data packet. Unicast and multicast on the other hand, are directed to specific targets; the other users on the network do not take any notice of the data or respond to it.

Unicast and multicast differ the most obviously in that unicast is only sent to an in­di­vid­ual recipient, while a whole group of targets can be addressed by a multicast. But the multicast is not simply a col­lec­tion of in­di­vid­ual unicasts. Instead messages are sent to a special multicast address. This causes redi­rec­tion via a router or server to all members of the multicast group. This group has to exist prior to data transfer and cannot be created at the time of sending from the source node.

The advantage of the multicast in contrast to the multiple unicast – i.e. the transfer to several unicast addresses – is that the data packet is only sent once. If you use several unicasts, the package is resent to the network each time. When sending data via multicast, it is only mul­ti­plied on the dis­tri­b­u­tion list – this saves bandwidth. Mul­ti­cas­ts are therefore par­tic­u­lar­ly popular for mul­ti­me­dia streaming. As a large group of clients requires exactly the same data, it is much more practical to only send it once. If the same data were to be sent over unicast to many re­cip­i­ents at the same time, it would be slower.