Link Efficiency Mechanisms
Although QoS mechanisms cannot actually create bandwidth, they can help
your network use the available bandwidth more efficiently. Two ways of
doing this are compression and fragmentation. These mechanisms are typically
applied at the WAN edge, where links are slower than within the LAN.
Compression
Compressing the traffic on a line creates more useable bandwidth; because
each frame is smaller, there are fewer bits to transmit. Thus, the serialization
delay is reduced, and more frames can be sent. Cisco IOS supports two
types of compression: payload and header.
Payload compression can be done at either Layer 2 or Layer 3.
■ Layer 2 Payload Compression—The Layer 2 payload compression
compresses the Layer 3 and 4 headers and the packet data. Layer 2
payload compression is typically a hop-by-hop mechanism, because
the Layer 2 header is removed at each hop. Layer 2 compression done
in software is CPU-intensive and might actually add extra delay to the
traffic flow. Hardware compression, however, adds little delay. Cisco
supports three Layer 2 payload compression algorithms:
— Stacker
— Predictor
— Microsoft Point-to-Point Compression (MPPC)
■ Layer 3 Payload Compression—Layer 3 payload compression
compresses the Layer 4 header and the packet data. It is generally
done session-by-session.
Header compression leaves the payload intact but compresses the headers.
TCP header compression compresses the IP and TCP headers. RTP header
compression compresses the IP, UDP, and RTP headers. It is most effective
when the headers are much larger than the payload, such as with Telnet or
VoIP. Headers do not change much over the life of a flow and contain many
redundant fields (such as source and destination IP address, protocol, and
port). Compression removes the redundant information and sends only the
new information and an index pointing to the unchanged information.