A closed-loop queueing model of flow-based label switches, supporting label reservation protocols of different label-setup and release policies, is presented. This model can emulate the behavior of TCP under the label switch when the maximum window size has been achieved and the packet loss rate is negligible. The label-setup policy is that the IP controller does not start to set up a label until the accumulated packets of the same flow in the switch buffer have exceeded a triggering threshold. Meanwhile, the reserved bandwidth is released when the flow is detected idle and the label-release timer has expired. This policy can achieve higher channel utilization with minimal label processing overhead in spite of suffering from certain delay penalty. To avoid unnecessary TCP timeout or large packet delay under such policy, we also introduce a label-setup timer. Norton's theorem is applied to obtain approximate solutions of this queueing model. Although the analytical method is an approximate one, the simulation results show that the accuracy is high and this model can clearly illustrate how the label-setup and the lable-release timer affect the system performance. Besides, one can observe the trade-off between the throughput and the channel utilization.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Ling-Chih KAO, Zsehong TSAI, "Performance Analysis of Flow-Based Label Switching: The Single IP Flow Model" in IEICE TRANSACTIONS on Communications,
vol. E83-B, no. 7, pp. 1417-1425, July 2000, doi: .
Abstract: A closed-loop queueing model of flow-based label switches, supporting label reservation protocols of different label-setup and release policies, is presented. This model can emulate the behavior of TCP under the label switch when the maximum window size has been achieved and the packet loss rate is negligible. The label-setup policy is that the IP controller does not start to set up a label until the accumulated packets of the same flow in the switch buffer have exceeded a triggering threshold. Meanwhile, the reserved bandwidth is released when the flow is detected idle and the label-release timer has expired. This policy can achieve higher channel utilization with minimal label processing overhead in spite of suffering from certain delay penalty. To avoid unnecessary TCP timeout or large packet delay under such policy, we also introduce a label-setup timer. Norton's theorem is applied to obtain approximate solutions of this queueing model. Although the analytical method is an approximate one, the simulation results show that the accuracy is high and this model can clearly illustrate how the label-setup and the lable-release timer affect the system performance. Besides, one can observe the trade-off between the throughput and the channel utilization.
URL: https://global.ieice.org/en_transactions/communications/10.1587/e83-b_7_1417/_p
Copy
@ARTICLE{e83-b_7_1417,
author={Ling-Chih KAO, Zsehong TSAI, },
journal={IEICE TRANSACTIONS on Communications},
title={Performance Analysis of Flow-Based Label Switching: The Single IP Flow Model},
year={2000},
volume={E83-B},
number={7},
pages={1417-1425},
abstract={A closed-loop queueing model of flow-based label switches, supporting label reservation protocols of different label-setup and release policies, is presented. This model can emulate the behavior of TCP under the label switch when the maximum window size has been achieved and the packet loss rate is negligible. The label-setup policy is that the IP controller does not start to set up a label until the accumulated packets of the same flow in the switch buffer have exceeded a triggering threshold. Meanwhile, the reserved bandwidth is released when the flow is detected idle and the label-release timer has expired. This policy can achieve higher channel utilization with minimal label processing overhead in spite of suffering from certain delay penalty. To avoid unnecessary TCP timeout or large packet delay under such policy, we also introduce a label-setup timer. Norton's theorem is applied to obtain approximate solutions of this queueing model. Although the analytical method is an approximate one, the simulation results show that the accuracy is high and this model can clearly illustrate how the label-setup and the lable-release timer affect the system performance. Besides, one can observe the trade-off between the throughput and the channel utilization.},
keywords={},
doi={},
ISSN={},
month={July},}
Copy
TY - JOUR
TI - Performance Analysis of Flow-Based Label Switching: The Single IP Flow Model
T2 - IEICE TRANSACTIONS on Communications
SP - 1417
EP - 1425
AU - Ling-Chih KAO
AU - Zsehong TSAI
PY - 2000
DO -
JO - IEICE TRANSACTIONS on Communications
SN -
VL - E83-B
IS - 7
JA - IEICE TRANSACTIONS on Communications
Y1 - July 2000
AB - A closed-loop queueing model of flow-based label switches, supporting label reservation protocols of different label-setup and release policies, is presented. This model can emulate the behavior of TCP under the label switch when the maximum window size has been achieved and the packet loss rate is negligible. The label-setup policy is that the IP controller does not start to set up a label until the accumulated packets of the same flow in the switch buffer have exceeded a triggering threshold. Meanwhile, the reserved bandwidth is released when the flow is detected idle and the label-release timer has expired. This policy can achieve higher channel utilization with minimal label processing overhead in spite of suffering from certain delay penalty. To avoid unnecessary TCP timeout or large packet delay under such policy, we also introduce a label-setup timer. Norton's theorem is applied to obtain approximate solutions of this queueing model. Although the analytical method is an approximate one, the simulation results show that the accuracy is high and this model can clearly illustrate how the label-setup and the lable-release timer affect the system performance. Besides, one can observe the trade-off between the throughput and the channel utilization.
ER -