Modeling link throughput in IP networks
View Patent ↗A model is given for performance evaluation of IP computer networks that are dominated by congestion-controlled traffic. The model includes heterogeneous TCP flows, UDP flows, short-lived TCP flows, and TCP flows in a differentiated services network. The performance of large-scale networks is estimated where flows may encounter multiple congested links.
1. A method of modeling heterogeneous congestion-responsive flows in a computer network having a plurality of links, the method comprising:
for a given link having a plurality of congestion-responsive flows of varying traffic characteristics,
(i) for each individual flow having specific traffic characteristics, determining a link throughput for the individual flow as a function of packet drop probability and flow round trip time, and
(ii) determining total flow throughput of the given link by summing the link throughputs of each of the individual flows.
2. A method according to claim 1 , wherein the varying traffic characteristics include packet size.
3. A method according to claim 1 , wherein the varying traffic characteristics include flow window size.
4. A method according to claim 1 , wherein the varying traffic characteristics include flow round trip time.
5. A method according to claim 4 , wherein for each flow, the round trip time is the greater of system time resolution and actual roundtrip time for the flow.
6. A method according to claim 1 , wherein the varying traffic characteristics include average number of packets acknowledged by an acknowledgement packet received by a sender of the flow from a receiver of the flow.
7. A method according to claim 6 , wherein for each flow, the average number of packets acknowledged is a ratio of a total number of packets to the product of a total number of acknowledgement packets times a term adjusting for a probability of an acknowledgement packet being dropped.
8. A method according to claim 1 , wherein the plurality of congestion-responsive flows include a Transmission Control Protocol (TCP) flow.
9. A method according to claim 8 , wherein the given link further has a plurality of constant bit rate flows and determining total flow throughput of the given link includes summing each of the constant bit rate flows.
10. A method according to claim 9 , wherein the constant bit rate flows include a User Datagram Protocol (UDP) flow.
11. A method according to claim 8 , wherein the TCP flow is a short-lived TCP flow.
12. A method according to claim 8 , wherein the network is a differentiated services network.
13. A method according to claim 1 , wherein the network is an Internet Protocol (IP) network.
14. A method according to claim 1 , wherein the given link includes a Random Early Detection (RED) router.
15. A method according to claim 1 , further comprising:
representing the total flow throughput as a function of packet drop probability and the round trip times of the individual flows.
16. A method according to claim 1 , further comprising:
representing the total flow throughput as a function of packet drop probability and average queue size.
17. A method according to claim 1 , further comprising:
determining a link equilibrium operating point.
18. A method according to claim 16 , wherein the link equilibrium operating point is determined as an intersection between a function of packet drop probability that represents average queue size and a function of average queue size that represents packet drop probability.
19. A method according to claim 1 , wherein the packet drop probability represents a random early detection (RED) drop function of average queue size for the given link, and the flow round trip time represents a sum of average queue size for the given link and summed propagation delay experienced by the given flow.
20. A method according to claim 1 , wherein the packet drop probability represents a term reflecting a random early detection (RED) drop function of average queue size for each link in the path of the individual flow, and the flow round trip time represents a sum of summed average queue size for each link in the path of the individual flow and summed propagation delay experienced by the given flow.
21. A link flow module for modeling heterogeneous congestion-responsive flows in a computer network having a plurality of links, the module comprising:
a link throughput component that determines total flow throughput for a given link having a plurality of congestion-responsive flows of varying traffic characteristics, the link throughput component including:
(i) an individual flow unit that, for each individual flow having specific traffic characteristics, determines a link throughput for the individual flow as a function of packet drop probability and flow round trip time, and
(ii) a flow summing unit, in communication with the individual flow unit, that determines total flow throughput of the given link by summing the link throughputs of each of the individual flows.
22. A module according to claim 21 , wherein the varying traffic characteristics include packet size.
23. A module according to claim 21 , wherein the varying traffic characteristics include flow window size.
24. A module according to claim 21 , wherein the varying traffic characteristics include flow round trip time.
25. A module according to claim 24 , wherein for each flow, the round trip time is the greater of system time resolution and actual roundtrip time for the flow.
26. A module according to claim 21 , wherein the varying traffic characteristics include average number of packets acknowledged by an acknowledgement packet received by a sender of the flow from a receiver of the flow.
27. A module according to claim 26 , wherein for each flow, the average number of packets acknowledged is a ratio of a total number of packets to the product of a total number of acknowledgement packets times a term adjusting for a probability of an acknowledgement packet being dropped.
28. A model according to claim 21 , wherein the plurality of congestion-responsive flows include a Transmission Control Protocol (TCP) flow.
29. A module according to claim 28 , wherein the given link further has a plurality of constant bit rate flows and the flow summing unit further sums each of the constant bit rate flows to determine total flow throughput of the given link.
30. A module according to claim 29 , wherein the constant bit rate flows include a User Datagram Protocol (UDP) flow.
31. A module according to claim 28 , wherein the TCP flow is a short-lived TCP flow.
32. A module according to claim 28 , wherein the network is a differentiated services network.
33. A module according to claim 21 , wherein the network is an Internet Protocol (IP) network.
34. A module according to claim 21 , wherein the given link includes a Random Early Detection (RED) router.
35. A module according to claim 21 , wherein the link throughput component further represents the total flow throughput for the given link as a function of packet drop probability and the round trip times of the individual flows.
36. A module according to claim 21 , wherein the link throughput component further represents the total flow throughput for the given link as a function of packet drop probability and average queue size.
37. A module according to claim 21 , wherein the link throughput component further determines a link equilibrium operating point.
38. A module according to claim 37 , wherein the link throughput component determines the link equilibrium operating point as an intersection between a function of packet drop probability that represents average queue size and a function of average queue size that represents packet drop probability.
39. A module according to claim 21 , wherein the packet drop probability represents a random early detection (RED) drop function of average queue size for the given link, and the flow round trip time represents a sum of average queue size for the given link and summed propagation delay experienced by the given flow.
40. A module according to claim 21 , wherein the packet drop probability represents a term reflecting a random early detection (RED) drop function of average queue size for each link in the path of the individual flow, and the flow round trip time represents a sum of summed average queue size for each link in the path of the individual flow and summed propagation delay experienced by the given flow.
41. A computer program product for use on a computer system for modeling heterogeneous congestion-responsive flows in a computer network having a plurality of links, the computer program product comprising a computer usable medium having computer readable program code thereon, the computer readable program code comprising:
program code for determining total flow throughput for a given link having a plurality of congestion-responsive flows of varying traffic characteristics, including
(i) program code for determining, for each individual flow having specific traffic characteristics, a link throughput for the individual flow as a function of packet drop probability and flow round trip time, and
(ii) program code for determining total flow throughput of the given link by summing the link throughputs of each of the individual flows.
42. A computer program product according to claim 41 , wherein the varying traffic characteristics include packet size.
43. A computer program product according to claim 41 , wherein the varying traffic characteristics include flow window size.
44. A computer program product according to claim 41 , wherein the varying traffic characteristics include flow round trip time.
45. A computer program product according to claim 44 , wherein for each flow, the round trip time is the greater of system time resolution and actual roundtrip time for the flow.
46. A computer program product according to claim 41 , wherein the varying traffic characteristics include average number of packets acknowledged by an acknowledgement packet received by a sender of the flow from a receiver of the flow.
47. A computer program product according to claim 46 , wherein for each flow, the average number of packets acknowledged is a ratio of a total number of packets to the product of a total number of acknowledgement packets times a term adjusting for a probability of an acknowledgement packet being dropped.
48. A computer program product according to claim 41 , wherein the plurality of congestion-responsive flows include a Transmission Control Protocol (TCP) flow.
49. A computer program product according to claim 48 , wherein the given link further has a plurality of constant bit rate flows and the program code for determining total flow throughput of the given link includes program code for summing each of the constant bit rate flows.
50. A computer program product according to claim 49 , wherein the constant bit rate flows include a User Datagram Protocol (UDP) flow.
51. A computer program product according to claim 48 , wherein the TCP flow is a short-lived TCP flow.
52. A computer program product according to claim 48 , wherein the network is a differentiated services network.
53. A computer program product according to claim 41 , wherein the network is an Internet Protocol (IP) network.
54. A computer program product according to claim 41 , wherein the given link includes a Random Early Detection (RED) router.
55. A computer program product according to claim 41 , further comprising:
program code for representing the total flow throughput as a function of packet drop probability and the round trip times of the individual flows.
56. A computer program product according to claim 41 , further comprising:
program code for representing the total flow throughput as a function of packet drop probability and average queue size.
57. A computer program product according to claim 41 , further comprising:
program code for determining a link equilibrium operating point.
58. A computer program product according to claim 57 , wherein the link equilibrium operating point is determined as an intersection between a function of packet drop probability that represents average queue size and a function of average queue size that represents packet drop probability.
59. A computer program product according to claim 41 , wherein the packet drop probability represents a random early detection (RED) drop function of average queue size for the given link, and the flow round trip time represents a sum of average queue size for the given link and summed propagation delay experienced by the given flow.
60. A computer program product according to claim 41 , wherein the packet drop probability represents a term reflecting a random early detection (RED) drop function of average queue size for each link in the path of the individual flow, and the flow round trip time represents a sum of summed average queue size for each link in the path of the individual flow and summed propagation delay experienced by the given flow.
61. A link flow module for modeling heterogeneous congestion-responsive flows in a computer network having a plurality of links, the module comprising:
means for determining total flow throughput for a given link having a plurality of congestion-responsive flows of varying traffic characteristics, the means for determining including:
(i) means, for each individual flow having specific traffic characteristics, for determining a link throughput for the individual flow as a function of packet drop probability and flow round trip time, and
(ii) means for summing the link throughputs of each of the individual flows to determine total flow throughput of the given link.
62. A link flow module according to claim 61 , wherein the varying traffic characteristics include packet size.
63. A link flow module according to claim 61 , wherein the varying traffic characteristics include flow window size.
64. A link flow module according to claim 61 , wherein the varying traffic characteristics include flow round trip time.
65. A link flow module according to claim 64 , wherein for each flow, the round trip time is the greater of system time resolution and actual roundtrip time for the flow.
66. A link flow module according to claim 61 , wherein the varying traffic characteristics include average number of packets acknowledged by an acknowledgement packet received by a sender of the flow from a receiver of the flow.
67. A link flow module according to claim 66 , wherein for each flow, the average number of packets acknowledged is a ratio of a total number of packets to the product of a total number of acknowledgement packets times a term adjusting for a probability of an acknowledgement packet being dropped.
68. A link flow module according to claim 61 , wherein the plurality of congestion-responsive flows include a Transmission Control Protocol (TCP) flow.
69. A link flow module according to claim 68 , wherein the given link further has a plurality of constant bit rate flows and means for determining total flow throughput of the given link includes means for summing each of the constant bit rate flows.
70. A link flow module according to claim 69 , wherein the constant bit rate flows include a User Datagram Protocol (UDP) flow.
71. A link flow module according to claim 68 , wherein the TCP flow is a short-lived TCP flow.
72. A link flow module according to claim 68 , wherein the network is a differentiated services network.
73. A link flow module according to claim 61 , wherein the network is an Internet Protocol (IP) network.
74. A link flow module according to claim 61 , wherein the given link includes a Random Early Detection (RED) router.
75. A link flow module according to claim 61 , further comprising:
means for representing the total flow throughput as a function of packet drop probability and the round trip times of the individual flows.
76. A link flow module according to claim 61 , further comprising:
means for representing the total flow throughput as a function of packet drop probability and average queue size.
77. A link flow module according to claim 61 , further comprising:
means for determining a link equilibrium operating point.
78. A link flow module according to claim 77 , wherein the link equilibrium operating point is determined as an intersection between a function of packet drop probability that represents average queue size and a function of average queue size that represents packet drop probability.
79. A link flow module according to claim 61 , wherein the packet drop probability represents a random early detection (RED) drop function of average queue size for the given link, and the flow round trip time represents a sum of average queue size for the given link and summed propagation delay experienced by the given flow.
80. A link flow module according to claim 61 , wherein the packet drop probability represents a term reflecting a random early detection (RED) drop function of average queue size for each link in the path of the individual flow, and the flow round trip time represents a sum of summed average queue size for each link in the path of the individual flow and summed propagation delay experienced by the given flow.