User manual

Packet Fragmentation
Packet fragmentation means splitting a data packet into several smaller packets.
We need to fragment packets because of:
o Hardware limits – some hardware do not support packets up to certain
threshold.
o Operating system buffer constraints – depending on the system memory,
buffer overflow means lost of data and waste bandwidth in unnecessary
retransmissions. Breaking packets into smaller segment means system will
have extra memory and extra time in processing those smaller data
packets.
o Protocol limits – some protocol specify maximum permitted size of a packet.
E.g. an ATM cell packet is restricted to 53 bytes.
o Reduce channel occupancy – routers can process smaller packets faster
than larger packets, result in smaller packet stays in router shorter time.
This will result in more throughputs and reduce the likelihood router
dropping packets.
Advantages of packet fragmentation:
 Easier and faster processing time for routers.
 Less delay compare to larger packets
 Less likelihood of routers dropping packets when traffic in network is heavy.
This is because the processing time for fragmented packets is lower.
Disadvantages for packet fragmentation:
Additional header information in the packet header result in less data can be
stored in a single packet.
Unnecessary packet fragmentation if traffic in network is low.
Reassemble fragmented packets requires extra time.
Setting fragmentation threshold can be a tricky business. Getting the best
performance out of fragmentation depends on the network traffic condition.
78