Packet Classification on SONET/SDH OC48/STM16 IQE PICs
Packet classification is used to partition the packets into different classes of traffic. You can use three methods to classify a packet:
Behavior aggregate (BA) classification
Fixed classification
Multifield classification
The SONET/SDH OC48/STM16 IQE PIC supports BA classification and fixed classification. It does not support multifield classification. However, multifield classification can be done at the Packet Forwarding Engine level using firewall filters, which overrides the classification done at the PIC level.
The BA classifier maps a class-of-service (CoS) value to a forwarding class and loss priority. The forwarding class determines the output queue. The loss priority is used by schedulers in conjunction with the weighted random early detection (WRED) algorithm to control packet discard during periods of congestion.
The SONET/SDH OC48/STM16 IQE PICs support the following BA classifiers:
DSCP IP or IP precedence
DSCP IPv6
MPLS (EXP)
The fixed classification matches the traffic on a logical interface level. The following example classifies all traffic on logical unit zero to the queue corresponding to assured forwarding.
[edit class-of-service interfaces so-0/1/2 unit 0] forwarding-class af;
If the classifiers are not defined explicitly, then the default classifiers are applied as follows:
All MPLS packets are classified using the MPLS (EXP) classifier. If there is no explicit MPLS (EXP) classifier, then the default MPLS (EXP) classifier is applied.
All IPv4 packets are classified using the IP precedence or DSCP classifier. If there is no explicit IP precedence or DSCP classifier, then the default IP precedence classifier is applied.
All IPv6 packets are classified using the DSCP IPv6 classifier. If there is no explicit DSCP IPv6 classifier, then the default DSCP IPv6 classifier is applied.