Intelligent switching of client packets among a group of servers

• US 7,937,490 B2
• Filed: 09/22/2008
• Issued: 05/03/2011
• Est. Priority Date: 06/18/2001
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: Open Invention Network LLC
• Inventors: Cheng, Bo-Chao, Wu, Tsong-Ho, Hsing, Deh-Phone K., Lu, Leonard L.
• Primary Examiner(s): Lim; Krisna
• Application Number:
  US12/235,367
• Publication Number:
  US 20090070489A1
• Time in Patent Office: 953 Days
• Field of Search: 709/245, 709/238, 709/236, 709/235, 370/355, 370/351, 370/389
• US Class Current: 709/235
• CPC Class Codes: H04L 45/74 Address processing for routing
• CPC Class Codes: H04L 47/10 Flow control; Congestion control
• CPC Class Codes: H04L 47/19 at layers above the network layer network arrangements for networked applications for scheduling or organising the servicing of application requests H04L67/60
• CPC Class Codes: H04L 47/2458 Modification of priorities while in transit
• CPC Class Codes: H04L 47/2466 using signalling traffic
• CPC Class Codes: H04L 47/25 with rate being modified by the source upon detecting a change of network conditions
• CPC Class Codes: H04L 47/32 by discarding or delaying data units, e.g. packets or frames
• CPC Class Codes: H04L 63/08 for authentication of entities cryptographic mechanisms or cryptographic arrangements for entity authentication H04L9/32
• CPC Class Codes: H04L 63/101 Access control lists [ACL]
• CPC Class Codes: H04L 67/02 based on web technology, e.g. hypertext transfer protocol [HTTP]
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/10015 Access to distributed or replicated servers, e.g. using brokers
• CPC Class Codes: H04L 67/1004 Server selection for load balancing
• CPC Class Codes: H04L 67/1008 based on parameters of servers, e.g. available memory or workload monitoring of computer activity G06F11/30
• CPC Class Codes: H04L 67/1014 based on the content of a request
• CPC Class Codes: H04L 67/1017 based on a round robin mechanism
• CPC Class Codes: H04L 67/1019 Random or heuristic server selection
• CPC Class Codes: H04L 69/22 Parsing or analysis of headers
• CPC Class Codes: H04L 9/40 Network security protocols

Abstract

The content-aware application switch and methods thereof intelligently switch client packets to one server among a group of servers in a server farm. The switch uses Layer 7 or application content parsed from a packet to help select the server and to schedule the transmitting of the packet to the server. This enables refined load-balancing and Quality of-Service control tailored to the application being switched. In another aspect of the invention, a slow-start server selection method assigned an initially boosted server load metric to a server newly added to the group of servers under load balancing. This alleviates the problem of the new server being swamped initially due to a very low load metric compared to that of others. In yet another aspect of the invention, a switching method dependent on Layer 7 content avoids delayed binding in a new TCP session. Layer 7 content is not available during the initial handshaking phase of a new TCP session. The method uses the Layer 7 content from a previous session as an estimate to help select the server and uses a default priority to scheduling the transmitting of the handshaking packets. Updated Layer 7 content available after the handshaking phase is then used to reset the priority for the transmit schedule and becomes available for use in load balancing of the next TCP session.

Claims

1. A method comprising:

a step for maintaining a server load metric for each server in a group of servers;

a step for parsing application content from a packet;

a step for selecting a destination server from the group of servers, the step for selecting the destination server being dependent on the server load metric for each server;

a step for assigning a priority to the packet, the priority being dependent on the application content; and

a step for transmitting the packet to the destination server according to a transmitting schedule, the transmitting schedule being dependent on the priority,

wherein the priority comprises a first priority or a second priority, the second priority is lower than the first priority, the destination server has a workload above a first redetermined level, and the transmitting schedule is constructed such that, if the packet comprises the first priority then the packet is transmitted to the destination server without delay and if the packet comprises the second priority then the packet is held back from being transmitted to the destination server.

2. The method of claim 1 further comprising:

determining at least one eligible server from among the group of servers, wherein determining at least one eligible server is dependent on the application content; and

a step for selecting the destination server from among the at least one eligible servers.

3. The method of claim 1 further comprising:

determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

wherein the step for selecting a destination server is also dependent at least in part upon the estimated application load.

4. The method of claim 1 wherein the transmitting schedule is also dependent at least in part upon the server load metric for each server.

5. The method of claim 1 wherein the application content comprises a Hypertext Transfer Protocol header.

6. A method comprising:

a step for maintaining a server load metric for each server in a group of servers;

a step for parsing application content from a packet;

determining at least one eligible server from among the group of servers, wherein determining at least one eligible server is dependent at least in part upon the application content;

a step for selecting a destination server from among the at least one eligible server;

determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

a step for assigning a priority to the packet, the priority being dependent at least in part upon the application content; and

a step for transmitting the packet to the destination server according to a transmitting schedule, the transmitting schedule being dependent at least in part upon the priority and the server load metric for each server;

wherein the step for selecting the destination server is dependent at least in part upon the server load metric for each server in the group of servers and the estimated application load.

7. The method of claim 6 wherein the application content comprises a Hypertext Transfer Protocol header.

8. The method of claim 6 wherein the priority comprises a first priority or a second priority, the second priority is lower than the first priority, the destination server has a workload above a first predetermined level, and the transmitting schedule is constructed such that, if the packet comprises the first priority then the packet is transmitted to the destination server without delay and if the packet comprises the second priority then the packet is held back from being transmitted to the destination server.

9. The method of claim 8 wherein the transmitting schedule is further constructed such that if the destination server has a workload below the first predetermined level then the packet is transmitted to the destination server without delay.

10. A method comprising:

a step for maintaining a server load metric for each server in a group of servers, wherein each server load metric provides a number of current server connections;

a step for adding a load balancing server to the group of servers, the load balancing server having a load balancing server load metric that provides a number of current server connections for the load balancing server;

a step for comparing the load balancing server load metric with an average server load metric for the group of servers;

a step for determining a disparity between the load balancing server load metric and the average server load metric;

a step for multiplying a time-varying factor to the load balancing server load metric such that the disparity is substantially reduced to below a predetermined value;

a step for selecting a destination server from among the group of servers, the destination server comprising a destination server load metric; and

a step for transmitting a packet to the destination server.

11. The method of claim 10 further comprising:

a step for weighting the server load metric of each server with a server weight, wherein the server weight indicates the capacity capability of each server.

12. The method of claim 10 further comprising:

a step for parsing application content from a packet;

determining at least one eligible server from among the group of servers, wherein determining at least one eligible server is dependent at least in part upon the application content; and

a step for selecting the destination server from the at least one eligible server.

13. The method of claim 10 further comprising:

determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

wherein the destination server load metric is also dependent at least in part upon the estimated application load.

14. The method of claim 12 wherein the application content comprises a Hypertext Transfer Protocol header.

15. The method of claim 10 further comprising:

a step for parsing application content from a packet;

determining at least one eligible server from among the group of servers, wherein determining at least one eligible server is dependent at least in part upon the application content; and

determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

wherein the destination server load metric is also dependent at least in part upon the estimated application load.

16. The method of claim 10 wherein the time-varying factor decreases from a first value to unity and the first value is substantially above unity.

17. The method of claim 16 wherein the first value comprises 2k and k is an integer, the method further comprising a step for reducing k by a factor of two until 2k becomes unity.

18. A method comprising:

a step for maintaining a server load metric for each server in a group of servers, wherein each server load metric provides a number of current server connections;

a step for adding a load balancing server to the group of servers, the load balancing server comprising a load balancing server load metric that provides a number of current server connections for the load balancing server;

a step for comparing the load balancing server load metric with an average server load metric for the group of servers;

a step for determining a disparity between the load balancing server load metric and the average server load metric;

a step for multiplying the load balancing server load metric by a factor of 2k, k being an integer;

a step for reducing k by a factor of two until 2k becomes unity;

a step for selecting a destination server from among the group of servers, the destination server comprising a destination server load metric; and

a step for transmitting a packet to the destination server.

19. The method of claim 18 further comprising:

a step for weighting the server load metric of each server with a server weight, wherein the server weight indicates the capacity capability of each server.

20. The method of claim 18 further comprising:

a step for parsing application content from the packet;

determining at least one eligible server from among the group of servers, wherein determining at least one eligible server is dependent at least in part upon the application content; and

a step for selecting the destination server from among the at least one eligible server.

21. The method of claim 20 further comprising:

determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

wherein the destination server load metric is also dependent at least in part upon the estimated application load.

22. The method of claim 21 wherein the application content comprises a Hypertext Transfer Protocol header.

23. The method of claim 20 further comprising:

a step for parsing application content from the packet;

determining at least one eligible server from among a group of servers, wherein determining at least one eligible server is dependent at least in part upon the application content; and

a step for determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

wherein the destination server load metric is dependent at least in part upon the estimated application load.

24. The method of claim 23 wherein the application content comprises a Hypertext Transfer Protocol header.

25. A method comprising:

a step for maintaining a server load metric for each server in a group of servers;

a step for parsing application content from a packet;

a step for updating the server load metric for each server in the group of servers when a TOP session is not in an initial handshaking phase;

determining an estimated application load based upon the application content;

a step for selecting a destination server from among the group of servers, the step for selecting being dependent at least in part upon the server load metric for each server;

a step for assigning a priority to the packet, wherein, if the TCP session is not in an initial handshaking phase then the priority is dependent at least in part upon the application content and if the TCP session is in an initial handshaking phase then the priority comprises a default priority; and

a step for transmitting the packet to the destination server according to a transmitting schedule that is dependent at least in part upon the priority.

26. The method of claim 25 wherein the default priority is highest among a group of priority types.

27. The method of claim 25 further comprising:

determining at least one eligible server from among the group of servers, wherein determining at least one eligible server is dependent at least in part upon the application content; and

a step for selecting the destination server from among the at least one eligible server.

28. The method of claim 25 further comprising:

determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

wherein the step for selecting the destination server is also dependent at least in part upon the estimated application load.

29. The method of claim 25 wherein the transmitting schedule is also dependent at least in part upon the server load metric for each server.

30. The method of claim 25 wherein the application content comprises a Hypertext Transfer Protocol header.

31. The method of claim 26 wherein the application content comprises a Hypertext Transfer Protocol header.

32. The method of claim 27 wherein the application content comprises a Hypertext Transfer Protocol header.

33. The method of claim 28 wherein the application content comprises a Hypertext Transfer Protocol header.

34. The method of claim 29 wherein the application content comprises a Hypertext Transfer Protocol header.

35. The method of claim 25 further comprising:

determining at least one eligible server from among the group of servers, wherein determining at least one eligible server is dependent at least in part upon the application content; and

a step for selecting the destination server from among the at least one eligible server.

36. The method of claim 26 further comprising:

determining an estimated application load for the destination server, the estimated application load being dependent at least in part upon the application content;

wherein the step for selecting a destination server is also dependent at least in part upon the estimated application load.

37. The method of claim 35 wherein the transmitting schedule is also dependent at least in part upon the server load metric for each server.

38. The method of claim 35 wherein the application content comprise a Hypertext Transfer Protocol header.

39. The method of claim 36 wherein the application content comprise a Hypertext Transfer Protocol header.

40. The method of claim 37 wherein the application content comprise a Hypertext Transfer Protocol header.