System and method for load balancing

• US 7,739,384 B2
• Filed: 12/13/2001
• Issued: 06/15/2010
• Est. Priority Date: 12/14/2000
• Status: Expired due to Fees

• Current Assignee: Flash Networks, Ltd. (Constellation Software Incorporated)
• Original Assignee: Flash Networks, Ltd. (Constellation Software Incorporated)
• Inventors: Laviatan, Chava, Siev, Amnon
• Primary Examiner(s): Srivastava; Vivek
• Assistant Examiner(s): Liu; Lin
• Application Number:
  US10/450,239
• Publication Number:
  US 20040071087A1
• Time in Patent Office: 3,106 Days
• Field of Search: 709/226, 709/225
• US Class Current: 709/226
• CPC Class Codes: G06F 2209/505 Clust
• CPC Class Codes: H04L 61/00 Network arrangements, protocols or services for addressing or naming
• CPC Class Codes: H04L 61/10 of different types
• CPC Class Codes: H04L 61/59 using proxies for addressing
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/1006 with static server selection, e.g. the same server being selected for a specific client
• 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 67/101 based on network conditions
• CPC Class Codes: H04L 67/1029 using data related to the state of servers by a load balancer
• CPC Class Codes: H04L 67/1031 Controlling of the operation of servers by a load balancer, e.g. adding or removing servers that serve requests

Abstract

The present invention is a method and system of load balancing in a group of one or more servers connected to one or more subnetworks. Two or more independent servers are bound into a group, with one of the servers elected to serve as a leader. The leader acts as a load balancer for the group while the remaining servers act as slaves. This functionality eliminates the need for one or more dedicated load balancing devices and lowers the hardware requirements necessary for performing such load balancing.

Claims

1. A method for performing symmetrical load balancing in a packet-based network, said method comprising:

spoofing a router, said router being located between a group of two or more servers having at least one joint IP address and a network, such that a network side of the router is connected to the network and a server side of the router is connected to a first side of the group of two or more servers,

so that at least one remote client subnetwork, which is connected to a second side of the group of two or more servers in such a way that the router and the group of one or more servers is communicatively positioned between the remote client subnetwork and the network and the remote client subnetwork having remote clients of the group of one or more servers,

appears to be connected to the server side interface of the router to which said group of one or more servers is connected;

electing one server from the group of two or more servers as a leader of the group;

receiving a request for a new connection between the at least one remote client subnetwork and the network; and

the leader of the group selecting a server from the group of two or more servers for said request and forwarding said request to the selected server; and

wherein the operation of the group of two or more servers is transparent to a remote client of the at least one remote client subnetwork and to a node in the network and a connection between the remote client of the at least one remote client subnetwork and the node in the network is using the IP address of said remote client and the IP address of said node in the network; and

wherein the network is based on the Internet Protocol.

2. The method of claim 1, wherein said spoofing the router comprises configuring its subnets list with the remote client subnet, pretending that the remote clients are connecting to an interface of said router.

3. The method of claim 1, wherein said spoofing the router comprises configuring its IP mapping with one of the joint IP address as the next HOP to subnetwork, which do not comprises remote clients of said group of one or more servers.

4. The method of claim 1, wherein said electing one server as a leader of the group comprises entering a boot mode wherein each server executes a selection algorithm.

5. The method of claim 4, wherein executing a selection algorithm comprises broadcasting a first type message, the first type message indicating whether the server has the feature to be a leader and receiving one or more messages from other servers.

6. The method of claim 5, wherein if the one or more messages from other servers indicate they have the feature to be a leader, applying arbitration to select one leader.

7. The method of claim 6, wherein applying arbitration comprises receiving an Internet protocol address for each server indicating the feature to be a leader, and selecting a leader based on the value of the Internet protocol address.

8. The method of claim 7, further comprising maintaining the remaining servers capable of being a leader in a slave mode after a leader is selected.

9. The method of claim 7, wherein selecting a leader based on the value of the Internet protocol address comprises comparing the Internet protocol address from each server capable of being a leader, and making the server with the highest internet protocol address the leader.

10. The method of claim 1, further comprising the step of the elected leader of the group further comprises the server elected as a leader broadcasting a message indicating the server is an active leader.

11. The method of claim 1, wherein said receiving a request for a new connection between at least one of the subnetworks and the network is initiated from the networks with an IP address of a remote client comprises grabbing the request by the router, broadcasting an ARP request with the IP address of the remote client to said group one or more servers, responding of the leader to the ARP request and forwarding it to a selected server.

12. The method of claim 1, wherein the selected server establishing the connection with the appropriate remote customer and responding to future ARP requests, from the router, with the IP address of said remote customer with the physical address of the selected server itself.

13. The method of claim 1, wherein the network is the Intranet.

14. The method of claim 1, wherein the network is an Internet.

15. The method of claim 1, wherein the subnetworks are Internet.

16. A system for establishing symmetrical load balancing in a packet-based network of connection between one or more remote subnetworks and a network, the system comprising:

a subnetwork comprising one or more remote clients;

a group of two or more servers, which is an intermediate node between the remote subnetwork and the network, with one server selected as a leader and said leader is operative to forward any request for new connection between the one or more remote clients of said subnetwork and the network, to one of the servers of the group of two or more servers; and

a router that is connected between said group of one or more servers and the network, the router being configured such that at least one remote client appears to be connected directly to the same interface of the router that the group of two or more servers is connected and wherein the one or more remote clients are clients of said group of two or more servers;

wherein the operation of the group of two or more servers is transparent to a remote client of the at least one remote client and to a node in the network and wherein a connection between the remote client of the at least one remote client subnetwork and the node in the network is using the IP address of said remote client and the IP address of said node in the network; and

wherein the network is based on the Internet Protocol.

17. The system of claim 16 wherein said group of one or more servers appears as a remote client.

18. The system of claim 16 wherein at least one of the subnetworks is capable for tunneling communication;

wherein the tunneling is over IP.

19. The system of claim 16 wherein said group of servers has a joint IP addresses and each server has a private IP address.

20. The system of claim 19 wherein only the leader is responsible on the joint IP addresses.

21. The system of claim 19 wherein the selecting process is based on the private IP address of each server.

22. The system of claim 19 wherein each server is connecting to each subnetwork via a different interface.

23. The system of claim 19 further comprising a router that is connected between said one of the subnetworks and said group of one or more servers.

24. A method for performing symmetrical load balancing in a group of two or more servers having at least one joint IP address, wherein the group of two or more servers is located in an intermediate node between a packet-based network and at least one remote client subnetwork, wherein the intermediate node is associated with a router of which a first side of the router is connected to the packet-based network and a second side of the router is connected to a first side of the group of two or more servers, a second side of the group of two or more servers is connected to the at least one remote client subnetwork, said method comprising the steps of:

spoofing the router so that the at least one remote client subnetwork, which is connected to the second side of the group of two or more servers, appears to be connected to the second side of the router as the group of two or more servers is connected;

electing one server from the group of two or more servers as a leader of the group;

receiving a request for a new connection between the at least one remote client subnetwork and the network; and

selecting, by the leader of the group, a server from the group of two or more servers for handling said request and forwarding said request to the selected server,

wherein the operation of the group of two or more servers is transparent to a remote client of the at least one remote client and to a node in the network and the request for a new connection between the remote client of the at least one remote client subnetwork and the node in the network is using the IP address of said remote client and the IP address of said node in the network; and

wherein the network is based on the Internet Protocol.