Serialization Delays (Transmission Time for Link Speed [in ms])
Link
Speed 64 Bytes 256 Bytes 512 Bytes 1024 Bytes 1500 Bytes
64 8 32 64 128 188
128 4 16 32 64 94
192 3 11 21 43 63
256 2 8 16 32 47
320 2 6 13 26 38
384 1 5 11 21 31
448 1 5 9 18 27
512 1 4 8 16 23
576 1 4 7 14 21
640 1 3 6 13 19
704 1 3 6 12 17
768 1 3 5 11 16
www.syngress.com
SerializationDelay =
Continued
264 Chapter 8 • Advanced QoS for AVVID Environments
832 1 2 5 10 14
896 1 2 5 9 13
960 1 2 4 9 13
1024 1 2 4 8 12
1088 0 2 4 8 11
1152 0 2 4 7 10
1216 0 2 3 7 10
1280 0 2 3 6 9
1344 0 2 3 6 9
1408 0 1 3 6 9
1472 0 1 3 6 8
1536 0 1 3 5 8
Using a feature like LLQ or PQ can significantly reduce delays on real-time
traffic, but even with this enabled, the time a real-time packet may have to wait
for even one large packet to be transmitted could be large enough to add jitter to
the stream.What usually happens is that after the priority queue empties, a large
packet is started out the interface. Shortly after this, another packet comes into
the priority queue but now has to wait for the whole large packet to be transmitted.
Meanwhile, other priority packets queue up behind the first one at regular
intervals.When the packets finally go, they go in a little burst. For an
application like VoIP, the jitter buffer may have difficulty playing out all these
packets smoothly with the delays and the bursts without dropping a packet or
adding an unacceptably large amount of delay.
Link Fragmentation and Interleaving overcomes this by reducing the maximum
packet size of all packets over a serial link to a size small enough that no
single packet will significantly delay critical real-time data.These large packets
that are broken up can now be interleaved with the real-time packets. LFI is
superior to just changing the maximum transmission unit (MTU) size, because
with LFI, fragmented packets are put back together at the other end of the serial
www.syngress.com
Table 8.5 Continued
Link
Speed 64 Bytes 256 Bytes 512 Bytes 1024 Bytes 1500 Bytes
Advanced QoS for AVVID Environments • Chapter 8 265
link.With MTU fragmentation, the packets travel across the whole network to
their destination before being reassembled.This causes unnecessary traffic and
processor utilization caused by increased header information.