dslreports logo
Networking 101: Packet Switching

Anyone that has used the internet will have a basic knowledge that data is being transferred back and forth, or more properly termed download and upload between their device, the internet, and ultimately a server. However, they may not realize the technology of how the data is actually moving.

The way that data gets moved is via ‘packet switching,’ and forms the basis for digital data communication. This is a technique that breaks the data into pieces, better known as packets or datagrams, which follow a standard format for digital transmission. By dividing the data into smaller pieces, this facilitates for an efficient transfer of information.

The packet contains not only the data to be sent, but also routing information. This additional routing information has several components, including:

• The address of the destination

• The address for the source

• The total count of the packets

• The sequence number

This additional information then serves to reassemble the data at the destination device. By having the total count of the packets, and their order, it allows the receiving computer to reorganize the data into the correct order at the destination.

The system is designed so that the data can tolerate out-of-order delivery. This commonly occurs as the different packets end up taking different routes due to multiple possible pathways, and variable network congestion along the way with variable latency as the packets of data traverse multiple nodes, and are forwarded. Packet switching is connectionless, as there is not one dedicated pathway between the devices transmitting and receiving the data.

The alternate to connectionless packet switching is connection-oriented packet switching, which also gets termed virtual circuit switching. In connection-oriented packet switching the data is transmitted over a predefined route, in the correct order. For this type of communication, a bidirectional environment needs to be present as a handshake is required so that the end-to-end connection gets defined. This gets called a ‘reliable’ connection as it guarantees that all the data gets sent, and in the correct order.

However, both our LAN’s, and the Internet are connectionless networks, and therefore the Ethernet and IP are also connectionless. Here there is no requirement of a dedicated session between the sender and the receiver. While there is no guarantee of delivery on the level of the network, this allows maximization of the bandwidth. However, the connection-oriented services can now get handled via the TCP, or Transport Control Protocol, as well as other protocols including X.25, Frame Relay or Multiprotocol Label Switching (MPLS).

Sending these data packets across the network, and through the Internet makes the process susceptible to congestion as the network gets overloaded. This results in ISP’s implementing traffic shaping solutions to deal with this network congestion, and maintain a Quality of Service (QoS). Manipulating the data packets to avoid congestion is a process known as Packet Shaping, or Traffic Shaping, with the overall goal of maximizing the network experience for the majority of users.

Packet switching is a key component of our communications across digital networks. Feel free to discuss future directions, opinions and challenges for this important technology.

This article was contributed by the DSLReports.com community. If you'd like to receive payment for writing content like this for our front page, please drop us a line.

Most recommended from 20 comments



davidc502
join:2002-03-06
Mount Juliet, TN

3 edits

3 recommendations

davidc502

Member

Circuit Switched

I was thinking we've come a long way since circuit switched networks were in place.... It's been a long time since I've thought about it, but seem to remember the reason why we have Packet Switched networks today was due to the DOD.... Apparently, in a circuit switched network if you loose a switch along the path there was no redundancy, so one would completely loose connection. One would have to wait for the switch to be fixed if another path wasn't found, before communications could resume. However with Packet switched technology we can loose routers, and if multiple routes exist, the packets will simply be routed around the problem. So, with packet Switching the DOD got what it wanted by having redundancy built into the network.

See the government was good for something...

cheers,

pclover
join:2008-08-02
Santa Cruz, CA

2 recommendations

pclover

Member

Not detailed enough.

Why too brief and confusing for new users with too many open ends.

Really need to understand the OSI model to understand these concepts as well as stateful vs stateless protocols.

Nothing is designed to handle out-of-order delivery at the network level. TCP offers this. UDP does not. Again, This is why the OSI model is important. There is a HUGE difference between L3 and L4 on the OSI Model.
said by digitaldoc77:

The packet contains not only the data to be sent, but also routing information. This additional routing information has several components, including:

• The address of the destination

• The address for the source

• The total count of the packets

• The sequence number

Here you are mixing L3 and L4 of the OSI model. The sequence number is a feature of TCP has zero bearing on IP routing decisions.