Customer Service and Booking Hotline: +44 (0) 845 672 0175

Search

Traffic Management Techniques for Packet Networks

Regardless of the technology used, certain fundamental traffic management techniques can be applied to optimise utilisation and handle congestion in packet-switched networks...

Connection Admission Control is normally applied when control plane signalling requests resources for a connection (ATM, Frame Relay, MPLS) or a flow (IP using IntServ). Typically all nodes along a path for which a reservation is requested must ensure that they can fulfil this request without degrading the performance of other connections or flows already in place and agreed. If these guarantees cannot be met, then the new connection is refused, and an alternative route or reduced requirements proposed.

Classification is the process of determining for each packet entering a network or node which of several defined classes it belongs to. Classification may be based on existing QoS markings, or on more general parameters of the packet, such as source address, destination address, or application being carried.

Marking is the process of placing a QoS value into the packet header, so that downstream processes n this node or other nodes can identify the QoS of the packet correctly and handle its queuing/scheduling and discarding appropriately

Policing is normally carried out at the head-end of a connection or flow to ensure that the contracted traffic rate is not being exceeded. Nodes may use shaping to smooth excessive bursts of traffic, and may discard or remark traffic if necessary according to the policies of the contract which applies

Congestion Management deals with situations where more traffic than can be carried by a node is presented to the output ports. Various techniques can be used to actively manage congestion, including explicit notification of congestion to sources of traffic, and selective discarding of traffic from queues.

Queuing and Scheduling allows different handling of packets based upon their class of service as defined by header markings or other classification within a node. Simple priority scheduling services the most important queue until it is empty, then services the next priority queue in the same way. Most nodes operate some more sophisticated weighted queuing algorithm which ensures some bandwidth for all priority classes.

Bookmark this article

Share this article using the following sites:

Courses by category...

Glossary Search

Newsletter Sign-up

Our RSS Feeds...