Network proxy with asymmetric connection connectivity

• US 7,768,939 B1
• Filed: 01/02/2007
• Issued: 08/03/2010
• Est. Priority Date: 01/02/2007
• Status: Active Grant

• Current Assignee: Juniper Networks Incorporated
• Original Assignee: Juniper Networks Incorporated
• Inventors: Trivedi, Atul N., Tosh, James K.
• Primary Examiner(s): Patel; Jayanti K
• Assistant Examiner(s): CAMPBELL, MATTHEW T
• Application Number:
  US11/619,084
• Time in Patent Office: 1,309 Days
• Field of Search: 370/254, 370/395.4, 370/241, 370/252, 709/224, 709/228, 709/223
• US Class Current: 370/252
• CPC Class Codes: H04L 43/10 Active monitoring, e.g. heartbeat, ping or trace-route
• CPC Class Codes: H04L 43/103 with adaptive polling, i.e. dynamically adapting the polling rate

Abstract

Techniques are described in which a network device waits differing amounts of time for different network sockets before beginning processes to determine whether respective network connections from the network sockets have failed. An intermediate device may create a network socket for a network connection having a keep-alive wait time option set to a keep-alive wait time associated with a class of the network connection. If an amount of time specified by the keep-alive option of the socket passes after a last successful communication on the network connection, the socket may begin a process to determine whether the network connection has failed. If the intermediate device determines that the network connection has failed, the intermediate device may terminate the connection to free resources on the intermediate device allocated to the network connection.

Claims

1. A method comprising:

storing, with a network device, respective wait times for a plurality of classes of network connections for a network protocol, wherein each of the wait times defines a time to wait before initiating a process to determine whether a network connection for the respective classes of network connections has failed, and wherein at least two of the wait times for the plurality of classes of network connections specify different wait times;

collecting statistical information from current network connections of the network device;

calculating the wait times for the plurality of classes of network connections based on the statistical information;

automatically identifying a new class of network connections based on the statistical information by identifying a set of the current network connections having a substantially similar time until failure that is different from the wait times for the currently defined plurality of classes of network connections;

calculating a wait time for the new class of network connections based on the statistical information;

establishing a socket with the network device for a network connection to a second network device, wherein the network connection is associated with a first one of the plurality of classes of network connections; and

initiating the process to determine whether the network connection has failed when the wait time for the first one of the plurality of classes of network connections has passed since a last successful communication on the second network connection.

2. The method of claim 1, wherein the socket comprises a first socket and the network connection comprises a first network connection, the method further comprising:

establishing, with the network device, a second socket for a network connection to a third network device, wherein the network connection is associated with a second one of the plurality of classes of network connections for the network protocol; and

initiating a process to determine whether the second network connection has failed when the second amount of time has passed since a last successful communication on the second network connection.

3. The method of claim 1, wherein the wait times for at least one of the plurality of classes of network connections is specified as a ratio of the wait time for a different one of the classes of network connections.

4. The method of claim 1, further comprising:

defining the different classes of network connections for different sub-networks for source network addresses of the network connections; and

setting the wait times for each of the classes of network communications based on the sub-networks.

5. The method of claim 1, further comprising:

defining the different classes of network connections for different application-layer protocols; and

setting the wait times for each of the classes of network communications based on the different application-layer protocols.

6. The method of claim 1, wherein the method further comprises:

receiving input specifying the wait times for each of the plurality of classes of network connections.

7. The method of claim 6, wherein receiving input comprises:

receiving the input from a human administrator through an administrative interface.

8. The method of claim 1, wherein the method further comprises:

automatically merging two or more of the classes of network connections into fewer classes of network connections when the statistical information for the current network connections indicate the two or more classes of network connections have a substantially similar time until closure.

9. The method of claim 1, wherein the network device serves as a network proxy, the method further comprising:

defining a first one of the classes of network connections as a class of network connections between the proxy device and a set of client devices;

defining a second one of the classes of network connects as a class of network connections between the proxy device and a set of server devices; and

automatically setting the wait time for the class of network connections with the client devices lesser than the wait time for the class of network connections with the server devices.

10. The method of claim 1,

wherein the first one of the plurality of classes of network connections comprise network connections from client devices attempting to access a first network service;

and

wherein a second one of the plurality of classes of network connection comprise network connections from client devices attempting to access a second network service.

11. The method of claim 1, wherein the method further comprises:

receiving a request to establish a network connection; and

analyzing the request to determine to which of the plurality of classes of network connections the requested network connection belongs.

12. The method of claim 11, wherein analyzing the request comprises:

determining criteria based on at least one of a source address, a destination address, a source port, a destination port and a protocol of the request; and

selecting one of the classes of network connections that matches the determined criteria.

13. The method of claim 1, wherein the network protocol comprises a Transmission Control Protocol (TCP).

14. The method of claim 1, wherein initiating a process to determine whether the network connection has failed comprises:

after expiration of the wait time for the network connection, initiating a process to send a keep-alive message on the network connection; and

determining that the network connection has failed when the network device does not receive a response to the keep-alive message on the first network connection.

15. The method of claim 14,

wherein the keep-alive message is a final keep-alive segment;

and

wherein the method further comprises;

prior to sending the final keep-alive segment, sending a first keep-alive segment on the first network connection;

waiting an idle time specified for the first class of network connections, wherein the idle time defines a time between keep-alive messages; and

sending the final keep-alive segment when the network device did not receive a response to the first keep-alive segment while waiting the specified period.

16. A network device comprising:

a processor;

a socket creation module, executing on the processor, to establish a first socket of a network device to manage a first network connection that is a member of a first class of network connections in a network protocol and to establish a second socket of the network device to manage a second network connection that is a member of a second class of network connections in the network protocol;

a statistical collection module, executing on the processor, to collect statistical information about network connections of the network device; and

an optimization module, executing on the processor to calculate a network connection keep-alive wait time for the first class of network connections and a network connection keep-alive wait time for the second class of network connections based on the statistical information,

wherein the optimization module automatically identifies a new class of network connections based on the statistical information and calculates a network connection keep-alive wait time for the new class of network connections based on the statistical information,

wherein the first socket determines whether the first network connection is still active when an amount of time that has passed since a last successful communication on the first network connection exceeds the network connection keep-alive wait time for the first class of network connections,

wherein the second socket determines whether the second network connection is still active when an amount of time that has passed since a last successful communication on the second network connection exceeds the network connection keep-alive wait time for the second class of network connections, and

wherein the network connection keep-alive wait time of the first socket is different than the network connection keep-alive wait time for the second socket.

17. The network device of claim 16, wherein the network device further comprises an administrative interface to receive input on which the first time for the first class of network connection and the second time for the second class of network connection is based.

18. The network device of claim 16,

wherein the first class of network connections comprise network connections between a set of client devices and the network device;

wherein the second class of network connections comprise network connections between the network device and a set of server devices; and

wherein the network device serves as a network proxy.

19. The network device of claim 16,

wherein the first class of network connections comprise network connections from client devices attempting to access a first network service;

and

wherein the second class of network connection comprise network connection from client devices attempting to access a second network service.

20. The network device of claim 16, wherein the socket creation module receives a request to establish a network connection and identifies the requested network connection as being a member of the first class of network connections.

21. The network device of claim 16, wherein the socket creation module creates the second socket to communicate to a second device information received from the first network connection.

22. The network device of claim 16, wherein the network protocol comprises a Transmission Control Protocol.

23. The network device of claim 16, wherein the first network socket determines whether the first network connection is still active by sending a keep-alive message on the first network connection and by determining that the network connection is no longer active when the network device does not receive a response to the keep-alive message on the first network connection.

24. The network device of claim 23, wherein the keep-alive message is a final keep-alive message and wherein the first network socket determines whether the first network connection is still active by further sending a first keep-alive message on the first network connection, waiting a period specified for the first class of network connections, and sending the final keep-alive message when the network device did not receive a response to the first keep-alive message while waiting the specified period.