| 摘要 |
A method and apparatus for avoiding packet reordering in multiple-class, multiple-priority networks. The present invention provides a queue implementation technique that can be used in multiple-class, multiple-priority networks such as Differentiated Services networks to ensure that packets are serviced without reordering. The queue implementation technique maintains performance isolation between different classes under some scheduling disciplines and can identify scheduling disciplines which do not degrade the performance seen by the lower priority traffic classes. The system includes a first queue for receiving packets associated with a first traffic class, the packets being arranged in the first queue as in-profile packets and out-profile packets and a second queue for storing pointers associated with the out-profile packets, wherein the packets in the first queue comprise doubly-linking, and the pointers associated with the out-profile packets in the second queue point to corresponding out-profile packets in the first queue as well as to a previous and next element in the second queue.</PTEXT> |
