On condition that the backup path has been pre-established,a scheme named EB for MPLS-based recovery is present,which introduced Backpressure signaling mechanism to Bridge scheme in order to eliminate the packet reordering.
According to this a novel quickly re-router model based on MPLS is presented,which mace use of links between working path and backup path (reduced the chance of lost packet ) as model to make the traffic in shorter path and avoid grouping disorder and longer delay caused by Haskin model,and more loss of data packages Makam model caused was avoided, At the same time,decrease the reserved resources used by Brige model.
· A packet order keeping load balance architecture named LBFOK and its packet distribute algorithm PDFC are proposed and analyzed. The analysis results show that LBFOK with PDFC mechanism get the characteristic of load balancing and packet order keeping.
A new bridge message through expanding the MPLS messages was proposed. And according to this a novel quickly re-router model --bridge model based on MPLS is presented, which make use of links between working path and backup path (reduced the chance of reversed traffic flow) as bridge to makes the traffic in shorter path and avoid grouping disorder and the longer delay caused by Haskin model. At the same time, more loss of data packages Makam model caused was avoided. Better performance of quick re-routing of...
A new bridge message through expanding the MPLS messages was proposed. And according to this a novel quickly re-router model --bridge model based on MPLS is presented, which make use of links between working path and backup path (reduced the chance of reversed traffic flow) as bridge to makes the traffic in shorter path and avoid grouping disorder and the longer delay caused by Haskin model. At the same time, more loss of data packages Makam model caused was avoided. Better performance of quick re-routing of a bridge model was shown through simulation results of NS2.
Packet reordering is a common event, which may affect and even depress TCP performance. Previous researches have declared that packet reordering is caused from parallel process switching system. But the distribution of the packet reordering scale has not been proposed yet. we build a simple simulation module of parallel packet switching system with OPNET. From this module some meaningful simulation results and conclusions are acquired. All these results are more reliable and reasonable.
Although TCP was a reliable transport mechanism; packet reordering aff ected its performance. The fundamental causes of packet reordering wer e parallel packet switch system and high-level traffic load. A model of packet paral lel swi tching system was presented to simulate how packet reordering depresses TCP cong estion window growth and degrades its performance. The simulations show that th e misinterpreting of out-of-order delivery as packet loss is the primary cau se for TCP performance degradation.