Shared LANs and Hubs: How They Work and Their Limitations

Within a shared LAN a hub will connect all devices to a single collision domain at the physical layer of the OSI model. In a shared LAN all traffic in a collision domain is seen by all of the other devices in the collision domain, traffic is not forwarded or filtered, it is just gets flooded (broadcasted).

The shared LAN approach introduces security risks into your network along with performance issues, as all devices are in the same collision domain and bandwidth is not used efficiently.

Consider the following scenario:

If Device A sends data to device B in a shared LAN environment such as the one illustrated in the diagram above, device B will receive the data as intended however device C and device D will also receive the data although it is not specifically meant for them.

Due to the devices forming part of the same collision domain, data can be lost in transmission when two devices transmit data at the same time. If device A and device C transmit simultaneously then their signals will collide and both transmissions are lost. Now using some clever technology called CSMA/CD the devices will be able to possibly *prevent and manage collisions on the network. (note the asterisk)

How CSMA/CD works

1. Frame Assembly:
   • Sending devices will start by assembling their data into the format of a frame.
2. Carrier Sense:
   • After creating the frame the sending device will have to listen to the line to ensure that the transmission medium is idle.
   • Only once it is idle will the sending device start transmitting their data over the transmission medium.
3. Transmission and Collision Detection:
   • The device will transmit the first bit of the data and then then it will monitor for a collision signal, if all goes well the device will continue to transmit the rest of the frame.
   • If a collision is detected through a collision signal the device will follow the collision detection procedure, stopping the data transmission immediately.
4. Collision Handling:
   • When a collision occurs, the device sends a jam signal to inform all other devices on the collision domain.
   • The sending device will then wait for a random backoff period before attempting to resend the data.
     • A random backoff period helps to prevent devices from trying to then retransmit at the same time again.
5. Retries:
   • The process continues with the device retrying transmission until the frame is fully sent.

Issues with Shared LANs and CSMA/CD

CSMA/CD helps with collision management, but still does not change how a shared LAN works through broadcasting data out to all devices. It leads to several problems:

• Security Risks: In a shared LAN the data is flooded to all devices and unauthorized devices such as device D might be able to capture traffic that they are not supposed to see at all.
• Performance Degradation: Collisions make devices retransmit lost data, wasting time and bandwidth.
• Scalability issues: As more devices are added to the shared LAN the chance of collisions increase, leading to more delays.

Although CSMA/CD helps manage collisions in a shared LAN, since data is still flooded to all devices. This leads to security risks, performance issues, and poor scalability.

Switched LANs (bridged LANs): What are they and how do they work in comparison to shared LANs

Switched LANs break a single collision domain into multiple smaller collision domains, minimizing the chance of collisions. Switched LANs make intelligent forwarding decisions based on the contents of the forwarding table (bridge table). The Switched LAN approach improves over all areas of the Shared LAN approach, increasing security and performance all thanks to bridging mechanisms.

Conclusion

Shared LANs are not recommended, and newer Switched LANs will improve security and performance. In my next post I will be showcasing the Juniper Networks implementation of Bridging Mechanisms.

Post Series: Mastering Layer 2 Switching and VLANs in Juniper Networks