Enterprise & Campus LAN Switching Overview

Local Area Network (LAN) is a logical explanation of how big a network should be called as a local.

Definition

• For 2 or more computer or communicating devices which are in a room or in a floor, in a building or in a campus, if they are connected, they are said to be connected to a LAN.

Characteristics

• It should be capable of providing high bandwidth, high speed, high-capacity communication.
• Owner of the premises is the owner of the LAN. To connect 2 computers, we do not lease(rent) link from service provider.
• Cost of deploying LAN should be cheap. Equipment used in a LAN should be compact and powerful to provide high bandwidth or high capacity or high-speed communication.
• Services in a campus define the architect of the campus. Next generation campus wide network architecture should be converged architecture to carry voice, video, data i.e., triple line service over single converge infrastructure.

In a campus, if there are 40,000 employee and if the campus authority wants to host his 40,000 employees with all the 3 services, i.e., data, video, voice, then campus have to lay 3 networks separately.
Data network with the centre device has a switch connecting to all the computers.
Voice network with a centre device as a PBX connecting all computer.
To provide multi party video conferencing then they had to lay a network with a centre device called as MCU (Multimedia convergence unit), connecting all computer.
Drawback

• Cost of laying one network is 2 crores, then cost of laying 3 network will be 6 crores. Hence, cost for separate network will increase.
• For maintaining this 3 Network separately, the campus had to appoint 3 group of people. Maintenance operational cost will be increased.
• All the devices are in line. Consider voice service device i.e., PBX which handle 40,000 voice call simultaneously. To provide these 40,000 simultaneous calls, the device itself should be very powerful. 
• In IP communication all the computers are given IP address to identify the computer in a communication and the information is delivered in the form of packet.

• Size of the premise is the size of LAN. Biggest LAN is the campus wide LAN.
• Administrative control is centralized.

Converged Architecture

When we build a converged architecture, your converged architecture will be a data network. Wire which come out of a switch, will go to your phone, and from phone it will be connected to a PC. Which mean your phone has a small switch inside for connectivity purpose. In this converged architecture, all the communication will be happened at the network or in the IP network. In IP enabled network, when voice, video, data all these 3 services are delivered together then it is very important to understand that these services will be going together on the same wire, they should not mix up and spoil the quality of experience. So, it is very important to understand what impact will happen if these 3 services will work together. For this we must study these services one by one.

Data Services:

• Data service is a best effort service.
• This mean there is no guarantee after how much time it will reach or after how much re-transmission it will reach or after how much drop it will reach, there is only one guarantee that whenever the data will reach it will reach perfectly.

Real Time Services:

• Real Time services are voice and video services.
• Stored video or stored voice are referred as data
• So, what is Real time voice and video services. When we are talking on a phone or watching a live cricket match on the screen is a real time voice and video.
• In Real time voice and video, the most important is Quality of Experience. It cannot be compromised, whatever the user is experiencing, the quality of that should be maintain.
• Quality of Experience is controlled by three parameters:

Delay (latency)
Jitter (Inconsistent delay)
Drops

• This parameter cannot be compromised in Real time services, whether the communication is happening in Real world or IP world.

Real world
Communication happens in FDM and in TDM
Every service work at different frequency
IP world
Communication between two devices is on same frequency.
If sending capacity is 10 Mbps, then receiving/hearing capacity should be 10 Mbps

• Now, if sending capacity is 100 Mbps, then some part is for data, some it for voice and rest it is for video. In Real time communication, in IP enabled world, these 3 parameters cannot be compromised, and it should be taken care, so that user experience is not spoil.
• Now, let us understand voice and video service.

Voice Service:

• When we speak on IP phone, it is actually a analog voice. When analog voice come out of mouth, the phone will digitize it and create sample of voice. This sample will be packetized and will be send on a network.
• Voice packet are very small. Normally, they are 8 Kb – 64 Kb in size.
• So, suppose if one of these packets is drop, it will not make a huge impact on user experience.
• Moreover, the end device has voice concealment program, which try to conceal the drop of voice sample. It works by concealing or by making the copy of received voice packet or forthcoming voice packet and just concealing in between the drop packet and maintain the quality of voice.
• The number of drops which can be handle should be 1/10000 packet.
• In Voice, drop is allowed, which is not allowed is delay or jitter.
• Therefore, Voice is Delay Sensitive.

Video Service:

• Video packet are of 2 types:

High definition – 8.5 mb
Standard definition – 2.5 mb

• These files can be created per second
• So, these files cannot be sent in one shot, so they are fragmented and sent one after the other in sequence.
• More important in video, is sequencing of video frame. If there is no proper sequencing user experience will be lost.
• To control sequencing of frame, three types of frames are created by a conferencing kit. They are:

I frame            :         B frame       :        P frame
Index Frame       Bearer Frame           Padding Frame

• I frame contain actual indexing of one second video.
• Drops are not allowed in video communication. Now, what is allowed is delay.
• Therefore, Video is Drop Sensitive.

When we used all these 3 services together on a wire, then every equipment on my network should be intelligent,
This required intelligent should be classification and prioritization.
When these 3 packets are received by any device, it should first classify which is voice, video and data packet, and after classifying it should prioritized voice over video, and video over data to maintain and provide quality to be work over converged architecture.
This capability of classification and prioritization is called Quality of service (QoS), which is only available in IP enabled network. So, converged architecture should be IP enabled network.

Ethernet Technology

Technology facilitate communication, it does not communicate on your behalf. Somebody, have to use technology for communication, it does not communicate on his own.
GSM is a Voice communication Technology, when you dial a number, it searches a number which can be presence in any corner of earth and connect through it.
Technology has 2-level:

• Level-1 ---- Hardware
• Level-2 ---- Software/Logic/Protocol

LAN is Ethernet Technology
And Mostly, MAN and WAN are Serial Technology.

• Ethernet Technology work on the fundamental of CSMA/CD.
• IEEE has standardized these fundamental of ethernet and given a number, IEEE 802.3
• Cisco has deployed this standard in his switches and routers and named as ARPA.

Carrier Sense Multiple Access with Collision Detection

• CSMA/CD is basically used on half-duplex Ethernet technology for local area networking.

• CSMA/CD, a MAC process protocol, first senses for any transmissions from the other stations in the channel and starts transmitting only when the channel is clear to transmit.
• As soon as a station detects a collision, it stops transmission and sends a jam signal. It then waits for some time before retransmitting.
• Let’s understand the meaning of the individual component of CSMA/CD

CS – It stands for Carrier Sensing. It implies that before sending data, a station first senses the carrier. If the carrier is found free, then the station transmits data else it refrains.
MA – Stands for Multiple Access i.e., if there’s a channel, then there are many stations that are trying to access it.
CD – Stands for Collision Detection. It also guides to proceed in case of packet data collision.

• CSMA/CD procedure can be understood as a group discussion, where if the participants speak all at once then it will be very confusing, and the communication will not happen.
• Instead, for good communication, it is required that the participants speak one after another so that we can clearly understand the contribution of each participant in the discussion.
• Once a participant has finished talking, we should wait for a certain period to see if any other participant is speaking or not. One should start speaking only when no other participant has spoken. If another participant also speaks at the same time, then we should stop, wait, and try again after some time.
• Similar is the process of CSMA/CD, where the data packet transmission is only done when the data transmission medium is free. When various network devices try to share a data channel simultaneously, then it will encounter a data collision.
• The medium is continuously monitored to detect any data collision. When the medium is detected as free, the station should wait for a certain period before sending the data packet to avoid any chances of data collision.
• When no other station tries to send the data and there is no data collision detected, then the transmission of data is said to be successful.

Algorithm
The algorithm steps include:

• First, the station that wants to transmit the data senses the carrier as to whether it is busy or idle. If a carrier is found idle, then the transmission is carried out.
• The transmission station detects a collision, if any, using the condition: Tt >= 2 * Tp where Tt is the transmission delay and Tp is the propagation delay.
• The station releases the jam signal as soon as it detects a collision.
• After collision has occurred, the transmitting station stops transmitting and waits for some random amount of time called the ‘back-off time’. After this time, the station retransmits again.

How Does CSMA/CD Work?
To understand the working of CSMA/CD, let’s consider the following scenario.

• Check if the sender is ready for transmitting data packets.
• Check if the transmission link is idle?

Sender must keep on checking if the transmission link/medium is idle. 
For this, it continuously senses transmissions from other nodes. 
Sender sends dummy data on the link. 
If it does not receive any collision signal, this means the link is idle now. 
If it senses that the carrier is free and there are no collisions, it sends the data. 
Otherwise, it refrains from sending data.

• Transmit the data & check for collisions.

Sender transmits its data on the link. CSMA/CD does not use an ‘acknowledgment’ system. 

It checks for successful and unsuccessful transmissions through collision signals. 

During transmission, if a collision signal is received by the node, transmission is stopped. 

The station then transmits a jam signal onto the link and waits for random time intervals before it resends the frame. 

After some random time, it again attempts to transfer the data and repeats the above process.

• If no collision was detected in propagation, the sender completes its frame transmission and resets the counters.

How does a station know if its data collide? 

Consider the above situation. Two stations, A & B.
Propagation Time: Tp = 1 hr (Signal takes 1 hr to go from A to B)

At time t=0, A transmits its data.
        t= 30 mins: Collision occurs.

After the collision occurs, a collision signal is generated and sent to both A & B to inform the stations about a collision. Since the collision happened midway, the collision signal also takes 30 minutes to reach A & B.

Therefore, t=1 hr: A & B receive collision signals.

This collision signal is received by all the stations on that link. Then,

How to ensure that it is our station’s data that collided?

• For this, Transmission time (Tt) > Propagation Time (Tp) [Rough bound]
• This is because, we want that before we transmit the last bit of our data from our station, we should at least be sure that some of the bits have already reached their destination. 
• This ensures that the link is not busy, and collisions will not occur.
• But above is a loose bound. We have not taken the time taken by the collision signal to travel back to us.

For this consider the worst-case scenario.
Consider the above system again.

At time t=0, A transmits its data.
        t= 59:59 mins: Collision occurs

This collision occurs just before the data reaches B. Now the collision signal takes 59:59 minutes again to reach A. Hence, A receives the collision information approximately after 2 hours, that is, after 2 * Tp. 

Hence, to ensure tighter bound, to detect the collision completely,
  Tt > >= 2 * Tp 

This is the maximum collision time that a system can take to detect if the collision was of its own data.
What should be the minimum length of the packet to be transmitted?
Transmission Time = Tt = Length of the packet/ Bandwidth of the link
[Number of bits transmitted by sender per second]
Substituting above, we get,
Length of the packet/ Bandwidth of the link>= 2 * Tp

Length of the packet >= 2 * Tp * Bandwidth of the link

Padding helps in the cases where we do not have such long packets. We can pad extra characters to the end of our data to satisfy the above condition.

Efficiency Of CSMA/CD

The efficiency of CSMA/CD is better than Pure ALOHA however there are some points that need to be kept in mind while measuring the efficiency of CSMA/CD. These include:

• If the distance increases, then the efficiency of CSMA/CD decreases.
• For Local Area Network (LAN), CSMA/CD works optimally but for long-distance networks like WAN, it’s not advisable to use CSMA/CD.
• If the length of the packet is bigger, then the efficiency increases but then again there is a limitation. The maximum limit for the length of the packets is 1500 bytes.

Advantages & Disadvantages Of CSMA/CD

Advantages

• Overhead is less in CSMA/CD.
• Whenever possible, it utilizes all the bandwidth.
• It detects collision within a very short span of time.
• Its efficiency is better than simple CSMA.
• It mostly avoids any kind of wasteful transmission.

Disadvantages

• Not suitable for large distance networks.
• Distance limitation is 2500 meters. Collision can’t be detected after this limit.
• Assignment of priorities cannot be done to certain nodes.
• As devices are added, the performance disrupts exponentially.

Applications of CSMA/CD

• CSMA/CD was used in shared media Ethernet variants(10BASE2,10BASE5) and in the early versions of twisted pair Ethernet that used repeater hubs.
• But nowadays, modern Ethernet networks are built with switches and full-duplex connections so that CSMA/CD is no longer used.
• In full-duplex mode, communication is possible in both directions. So there are least or in fact no chance of collision and thus no mechanism like CSMA/CD find its use on a full-duplex.