Top Ad unit 728 × 90

08: Network layer - Fragmentation, path MTU discovery, and reassembly part 3

[post_ad]
 Fragmentation :

The term clearly explains itself. Fragmentation at the Network layer means spliting packets up into pieces. You may wonder why fragmenting packets if data is already segmented at the Transport layer.  The answer is that things are totally different. Fragmentation happens while passing through different devices. That is, each device fragments a given packet depending on the MTU ( Maximum Transmission Unit) that the network supports. Different networks have different maximum packet sizes. The MTU of Ethernet is not similar to that of WiFi, thus packets will be fragmented according to the network characteristics. Consider the following figure. 
The above figure shows that Network 1 supports 2500 KB per packet, so when the packet leaves the network to the next adjacent network, here fragmentation is necessary because Network 2 supports only 1900 KB as MTU. Each packet passing by this network will be split up into small fragments to fit its MTU which is 1900 KB.  Furthermore, the packets are, of course, on their way to the host located on the left in the figure, and again they are obliged to cross Network 3 in order to reach the host, so they will once more be fragmented to fit MTU 900 KB of the network. 

All in all, packets are fragmented according to the network devices they travel through. MTU of each network is what decides for the packets’ size. It is significant to mention that Ip header is included in each fragment. 

However, fragmentation is undesirable on the grounds that it causes more work for routers and hosts, tends to magnify loss rate, and what is more is security vulnerabilities. That is why a new technique was developed ; it is to solve the the aforementioned drawbacks. Path MTU Discovery is what is used to avoid fragmentation. 

Path MTU Discovery :

This technique tests path with large packet and once it is found, the network MTU’s size is dependent on in each packet transmission. This method is widely employed today on the grounds of its efficiency; it replaces fragmentation. See the following figure to have it grasped.
The above pictured figure shows that PC 1 is communicating with PC 2. Once a packet gets out of PC1, it will definitely cross Network 1, 2, 3 to reach Network 4 wherein PC 2 is located. The process of Path MTU Discovery is as follow :
PC 1 sends the first packet to test  path in order to see what size is the largest for MTU. Once the packet crosses Network 2, the router immediately sends its feedback telling the host that 1500 KB is the MTU, then the packet proceeds with transmission to reach Network 3 in which the router provides feedback informing the host (PC1) that MTU is 1200 KB, after that the packet will continue its way to Network 4 wherein the router will let the host (PC1) know that the largest size is 800 KB. Finally, the host ( PC1) will set its MTU to 800, and all transmitting packets will be 800 KB until the last packet sent.


Reassembly :

This process is almost similar to the one takes place at the Transport layer. Packets may arrive at the destination in a disorderly way because they may take different routes. A given route may be faster than the other depending on the bandwidth, so packets taking a route with slow speed will undoubtedly make it to the destination behind the times. In this case, the destination is compelled to reassemble the packets (segments) into the proper order ; this process is referred to as Reassembly.

[post_ad]
08: Network layer - Fragmentation, path MTU discovery, and reassembly part 3 Reviewed by BOUFTIRA on 10:00:00 AM Rating: 5

No comments:

All Rights Reserved by Computer Networks For All © 2015 - 2017

Contact Form

Name

Email *

Message *

Powered by Blogger.