Global server load balancing

• US 7,454,500 B1
• Filed: 09/26/2000
• Issued: 11/18/2008
• Est. Priority Date: 09/26/2000
• Status: Expired due to Term

• Current Assignee: Avago Technologies International Sales Pte Limited (Broadcom, Inc.)
• Original Assignee: Foundry Networks Incorporated (Broadcom, Inc.)
• Inventors: Cheung, David Chun Ying, Hsu, Ivy Pei-Shan, Jalan, Rajkumar Ramniranjan
• Primary Examiner(s): SALAD, ABDULLAHI ELMI
• Application Number:
  US09/670,487
• Time in Patent Office: 2,975 Days
• Field of Search: 709/226, 709/245, 709/105, 709/201, 709/227, 709/228, 709223-224, 709/238, 709/203, 709/217, 709/219, 709/225, 709/241, 707/10, 370/389, 370/400, 370/401
• US Class Current: 709/226
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• 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 67/1012 based on compliance of requirements or conditions with available server resources
• CPC Class Codes: H04L 67/1021 based on client or server locations

Abstract

A global server load balancing (GSLB) switch serves as a proxy to an authoritative DNS communicates with numerous site switches which are coupled to host servers serving specific applications. The GSLB switch receives from site switches operational information regarding host servers within the site switches neighborhood. When a client program requests a resolution of a host name, the GSLB switch, acting as a proxy of an authoritative DNS, returns one or more ordered IP addresses for the host name. The IP addresses are ordered using metrics that include the information collected from the site switches. In one instance, the GSLB switch places the address that is deemed “best” at the top of the list.

Claims

1. A method of load balancing among host servers of a data network, the method comprising;

storing, in a load balancing switch of the data network, round trip time data for a plurality of host server site switches, wherein the round trip time data for a host server site switch from the plurality of host server site switches indicates a time for exchanging at least one message between the host server site switch and a first client machine of the data network, wherein each host server site switch from the plurality of host server site switches is associated with one or more host servers of the data network, the one or more host servers associated with a host server site switch being reachable via the host server site switch; and

ordering, in the load balancing switch, a plurality of network addresses, the plurality of network addresses being responsive to a query regarding a host name, the plurality of network addresses determined from resolution of the host name, the plurality of network addresses comprising network addresses of multiple host server site switches from the plurality of host server site switches, wherein the load balancing switch is capable of ordering the plurality of network addresses based, at least in part, on the round trip time data stored for the multiple host server site switches.

2. The method of claim 1, further comprising;

creating a table, in the load balancing switch, using the round trip time data.

3. The method of claim 2, wherein the table is indexed by network neighborhood.

4. The method of claim 1, further comprising;

sending a health check message to each of the plurality of network addresses from the load balancing switch.

5. The method of claim 4, wherein the health check is a layer 7 health check.

6. The method of claim 5, wherein the layer 7 health check uses a HTTP protocol or a FTP protocol.

7. The method of claim 4, wherein the health check is a layer 4 health check.

8. The method of claim 7, wherein the health check is a TCP or UDP health check.

9. The method of claim 1, wherein

the first client machine is one of a plurality of client machines of the data network;

and

the round trip time data for a host server site switch comprises time information for exchanging at least one message between the host server site switch and each client machine from the plurality of client machines with which the host server site switch establishes a connection.

10. The method of claim 1, wherein for a host server site switch from the plurality of host server site switches, the time for exchanging at least one message between the host server site switch and the first client machine is a time difference between the receipt, at the host server site switch, of a connection request message and a connection acknowledgement message from the first client machine.

11. The method of claim 10, wherein the connection request message comprises a TCP SYN packet and the connection acknowledgment message comprises an associated TCP ACK packet.

12. The method of claim 1, wherein the query originated at the first client machine;

and wherein the ordering comprises;

generating an ordered list of network addresses from a first network address from the plurality of network addresses to a last network address from the plurality of network addresses, wherein the first network address is associated with a host server site switch from the plurality of host server site switches having a lowest round trip time with the first client machine and the last network address is associated with a host server site switch from the plurality of host server site switches having a highest round trip time with the first client machine.

13. The method of claim 1, wherein the ordering comprises generating an ordered list of network addresses such that a network address from the plurality of network addresses associated with a host server site switch having a higher round trip time is ranked higher in the ordered list than a network address from the plurality of network addresses associated a host server site switch having a lower round trip time.

14. The method of claim 1 further comprising:

determining, based upon the ordering, if the plurality of network addresses comprises a single network address corresponding to a host server site switch with a round trip time data that is less than the round trip time data for any other host server site switch corresponding to a network address in the plurality of network addresses; and

upon determining that the plurality of network addresses does not comprise a single network address, reordering the plurality of network addresses using a criterion that is different than the round trip time data for the multiple host server site switches.

15. A method of load balancing among host servers of a data network, the method comprising;

receiving, at a load balancing switch of the data network, a query regarding a host name;

selecting, from a plurality of network addresses determined responsive to the query, a best network address based, at least in part, on which of the plurality of network addresses has been least recently selected by the load balancing switch as a best network address in response to previous queries; and

storing, at the load balancing switch, round trip time data received from a plurality of host server sites associated with the host servers, wherein the round trip time data received from a host server site comprises a time for exchanging at least one message between the host server site switch client machine from a plurality of client machines.

16. The method of claim 15, wherein the time for exchanging messages between the host server site switch and the client machine is a time difference between the receipt, at the host server site switch, of a connection request message and a connection acknowledgement message from the client machine.

17. The method of claim 15, further comprising:

creating, in the load balancing switch based on the round trip time data, a proximity table.

18. The method of claim 15, further comprising:

creating, in the load balancing switch, a proximity table.

19. A load balancing switch for load balancing amongst a plurality of host servers of a data network, the load balancing switch comprising:

a means for storing round trip time data received from a plurality of host server site switches, the round trip time data received from a host server site switch being a time for exchanging at least one message between the host server site switch and a first client machine from a plurality of client machines of the data network, wherein each host server site switch from the plurality of host server site switches is associated with one or more host servers from the plurality of host servers, the one or more host servers associated with a host server site switch being reachable via the host server site switch;

a means for receiving a query regarding a host name, the query originating at the first client machine of the data network; and

a means for ordering a plurality of network addresses that are responsive to the query based, at least in part, on the stored round trip time data for host server site switches from the plurality of host server site switches whose network addresses are included in the plurality of network addresses.

20. The load balancing switch of claim 19, further comprising:

a means for ordering the plurality of network addresses based, at least in part, on which of the network addresses in the plurality of network addresses has been least recently selected as a best network address in response to previous queries.

21. The load balancing switch of claim 19, further comprising:

a means for ordering the plurality of network addresses based, at least in part, on a session capacity associated with host server site switches from the plurality of host server site switches whose network addresses are included in the plurality of network addresses.

22. The load balancing switch of claim 19, further comprising:

a means for ordering the plurality of network addresses based, at least in part, on an available session capacity associated with host server site switches from the plurality of host server site switches whose network addresses are included in the plurality of network addresses, wherein the available session capacity associated with a host server site switch is a percentage of a session capacity of the host server site switch.

23. The load balancing switch of claim 19, further comprising:

a means for ordering the plurality of network addresses based, at least in part, on a health of the host servers.

24. The load balancing switch of claim 19 further comprising:

means for determining, based upon the ordering, if the plurality of network addresses comprises a single network address corresponding to a host server site switch with a round trip time data that is less than the round trip time data for any other host server site switch corresponding to a network address in the plurality of network addresses; and

upon determining that the plurality of network addresses does not comprise a single network address, means for reordering the plurality of network addresses using a criterion that is different than the round trip time data for the multiple host server site switches.

25. A load balancing switch comprising:

a database configured to store round trip time data for a plurality of host server site switches, the round trip time data for each host server site switch from the plurality of host server site switches indicating a time for exchanging at least one message between the host server site switch and a client machine, each host server site switch from the plurality of host server site switches being associated with one or more host servers, the one or more host servers associated with a host server site switch being reachable via the host server site switch; and

a module configured to order a plurality of network addresses, the plurality of network addresses received by the load balancing switch in response to a query originating at the client machine, the plurality network addresses comprising network addresses determined from resolving a host name identified in the query, the plurality of network addresses comprising network addresses of multiple host server site switches from the plurality of host server site switches, wherein the network addresses in the plurality of network addresses are ordered based upon the round trip time data stored in the database for the multiple host server site switches.

26. The load balancing switch of claim 25 wherein a table is created using the round trip time data for the plurality of host server site switches and indexed by network neighborhood.

27. The load balancing switch of claim 25 further comprising a health check module configured to send a health check message to each of the plurality of network addresses.

28. The load balancing switch of claim 27 wherein a health check sent to a network address in the plurality of network addresses is at least one of a layer 7 health check or a layer 4 health check.

29. The load balancing switch of claim 25, wherein the time for exchanging messages between a host server site switch and the client machine is a time difference between the receipt, at the host server site switch, of a connection request message and a connection acknowledgement message from the client machine.