Link Fragmentation and Interleaving
Classify
R
R
R R R
R
R
R
R
F
F
F
F F F F
F F
Small Real-Time
Packets
Large Packets
Fragmentation
Queuing
To Interface
How is the fragmentation size chosen? A particular packet size corresponds to
a serialization delay.We choose the serialization delay by considering the maximum
delay tolerated by the critical application. From this delay, the fragmentation
size can be calculated as the product of the link speed and the target delay.
Let us illustrate this with an example.
Imagine that we have a VoIP application running in the company of other
data on a 128 Kbps circuit. Ethernet has an MTU of 1500 bytes, so on a 128
www.syngress.com
Figure 8.6 Link Fragmentation and Interleaving
Classify
R
R
R R R
R
R
R
R
F
F
F
F F F F
F F
Small Real-Time
Packets
Large Packets
Fragmentation
Queuing
To Interface
Outgoing
Packets
Real-Time Packet
Fragment
F
Incoming
Traffic
266 Chapter 8 • Advanced QoS for AVVID Environments
Kbps circuit without LFI, it would take 94 milliseconds (ms) to serialize the
entire packet (refer back to Table 8.5).Therefore, a VoIP packet could potentially
wait 94 ms before it could begin to be transmitted.This delay is too long and
would cause jitter in the playout stream of the listener.VoIP is usually sent with
two 10-ms samples in each packet.Assume we want to set the target delay
between each packet to 10 ms.The fragmentation size for this circuit is thus calculated
to be 160 bytes (128 Kbps x 10 ms).Therefore, to guarantee the target
delay of 10 ms, each large packet needs to be fragmented into 160-byte pieces.