Method and system for using bayesian network inference for selection of transport protocol algorithm

• US 20080144663A1
• Filed: 12/14/2006
• Published: 06/19/2008
• Est. Priority Date: 12/14/2006
• Status: Active Grant

• Current Assignee: Oracle America, Inc. (Oracle Corporation)
• Original Assignee: Sun Microsystems Incorporated (Oracle Corporation)
• Inventors: Marcondes, Cesar A.C., Johnson, Darrin P., Persson, Anders D.
• Granted Patent: US 7,672,240 B2
• Time in Patent Office: Days
• US Class Current: 370/469
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: H04L 69/16 Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
• CPC Class Codes: H04L 69/161 Implementation details of TCP/IP or UDP/IP stack architecture; Specification of modified or new header fields
• CPC Class Codes: H04L 69/163 In-band adaptation of TCP data exchange; In-band control procedures
• CPC Class Codes: H04L 69/32 Architecture of open systems interconnection [OSI] 7-layer type protocol stacks, e.g. the interfaces between the data link level and the physical level
• CPC Class Codes: H04L 69/323 in the physical layer [OSI layer 1]
• CPC Class Codes: H04L 69/325 in the network layer [OSI layer 3], e.g. X.25 H04L69/16 takes precedence
• CPC Class Codes: H04L 69/326 in the transport layer [OSI layer 4] H04L69/16 takes precedence

Abstract

A method for optimizing a network stack includes inputting network information into a transport protocol algorithm selector, inputting a first transport protocol algorithm into the transport protocol algorithm selector, analyzing a result of the transport protocol algorithm selector, selecting the first transport protocol algorithm based on the result, receiving a first packet in the network stack, and processing the first packet using the first transport protocol algorithm.

Claims

1. A method for optimizing a network stack, comprising:

inputting network information into a transport protocol algorithm selector;

inputting a first transport protocol algorithm into the transport protocol algorithm selector;

analyzing a result of the transport protocol algorithm selector;

selecting the first transport protocol algorithm based on the result;

receiving a first packet in the network stack; and

processing the first packet using the first transport protocol algorithm.

2. The method of claim 1, further comprising:

mining a network characteristic from the network information using an inline estimator on the network stack;

generating an estimate of the network characteristic using the inline estimator; and

updating the transport protocol algorithm selector based on the estimate.

3. The method of claim 1, further comprising:

determining an optimality of the first transport protocol algorithm;

selecting a second transport protocol algorithm based on the optimality of the first transport protocol algorithm;

receiving a second packet in the network stack; and

processing the second packet using the second transport protocol algorithm.

4. The method of claim 3, further comprising:

resetting the transport protocol algorithm selector based on the optimality of the first transport protocol algorithm.

5. The method of claim 1, wherein the result is obtained using an anytime inference algorithm.

6. The method of claim 1, wherein the transport protocol algorithm selector comprises a Bayesian network.

7. The method of claim 6, wherein the Bayesian network comprises a plurality of nodes, wherein each of the nodes is one selected from a group consisting of the first transport protocol algorithm and one of a plurality of network conditions.

8. The method of claim 7, wherein the plurality of network conditions include at least one selected from a group consisting of a system load, a round trip time, a path bandwidth, a buffer size, a loss rate, and an available bandwidth.

9. The method of claim 6, wherein the result comprises a leaf node in a Bayesian network.

10. A method for optimizing a network stack, comprising:

inputting network information into a transport protocol algorithm selector;

inputting an expected result into the transport protocol algorithm selector;

inferring a first transport protocol algorithm that produces the expected result using the transport protocol algorithm selector;

receiving a first packet in the network stack; and

processing the first packet using the first transport protocol algorithm.

11. The method of claim 10, further comprising:

mining a network characteristic from the network information using an inline estimator on the network stack;

generating an estimate of the network characteristic using the inline estimator; and

updating the transport protocol algorithm selector based on the estimate.

12. The method of claim 10, wherein the transport protocol algorithm is inferred using an anytime inference algorithm.

13. The method of claim 10, wherein the transport protocol algorithm selector comprises a Bayesian network.

14. A system for selecting a transport protocol algorithm, comprising:

a network stack comprising network information;

a transport protocol algorithm selector, comprising;

a plurality of inputs corresponding to a plurality of network conditions based on the network information and the transport protocol algorithm; and

a plurality of outputs corresponding to a plurality of results,

wherein the plurality of results is used to select the transport protocol algorithm for use on the network stack.

15. The system of claim 14, further comprising:

an inline estimator on the network stack,

wherein the inline estimator is used to mine a network characteristic of the network stack and generate an estimate of the network characteristic, and

wherein the estimate is used to update the transport protocol algorithm selector.

16. The system of claim 14, wherein the plurality of results is obtained using an anytime inference algorithm.

17. The system of claim 14, wherein the inline estimator is used to determine an optimality of the first transport protocol algorithm, and wherein a second transport protocol algorithm is used on the network stack based on the optimality of the first transport protocol algorithm.

18. The system of claim 17, wherein the transport protocol algorithm selector is reset based on the optimality of the first transport protocol algorithm.

19. The system of claim 14, wherein the transport protocol algorithm selector comprises a Bayesian network.

20. The system of claim 19, wherein the plurality of inputs correspond to a plurality of root nodes in the Bayesian network, and wherein the plurality of outputs correspond to a plurality of leaf nodes in the Bayesian network.