CLUSTER CONTROL SYSTEM, CLUSTER CONTROL METHOD, AND PROGRAM

• US 20110231508A1
• Filed: 12/02/2009
• Published: 09/22/2011
• Est. Priority Date: 12/03/2008
• Status: Active Grant

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Torii, Takashi
• Granted Patent: US 8,782,160 B2
• Time in Patent Office: Days
• US Class Current: 709/208
• CPC Class Codes: H04L 61/103 across network layers, e.g. resolution of network layer into physical layer addresses or address resolution protocol [ARP]
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers

Abstract

A slave computer out of a plurality of slave computers that are connected to a plurality of clients through a front network and provide the front network with a single IP address that is a same IP address comprises: a responsible client table that indicates a client for which each of the plurality of slave computers is responsible; and an ARP processing unit that refers to the responsible client table when an ARP request for the single IP address has been received from one of the plurality of clients and sends an ARP response when a client that has sent the ARP request is present in the responsible client table, and otherwise does not send the ARP response, wherein a client in the responsible client table is added, changed, or deleted by a master computer connected through a back-end network.

Claims

1. A cluster control system comprising:

a plurality of clients;

a server group that is connected to each of the plurality of clients through a front network; and

a master computer that is connected to the server group through a back-end network and controls the server group, wherein

the server group comprises a plurality of slave computers that provide the front network with a single IP (Internet Protocol) address that is a same IP address;

each of the plurality of slave computers comprises;

a responsible client table that indicate a client for which each of the plurality of slave computers is responsible; and

an ARP processing unit that refers to the responsible client table when an ARP (Address Resolution Protocol) request for the single IP address has been received from each of the plurality of clients, sends an ARP response when a client that has sent the ARP request is present in the responsible client table, and does not send an ARP response when a client that has sent the ARP request is not present in the responsible client table; and

the master computer comprises a responsible node determination processing unit that maintains and manages through the back-end network in the responsible client table information regarding addition, change, or deletion of a client for which each of the plurality of slave computers is responsible.

2. The cluster control system according to claim 1, wherein

each of the plurality of slave computers further comprises:

an ARP notification unit that, when receiving an ARP request from a client that is not present in the responsible client table, notifies to the master computer an IP address and a MAC (Media Access Control) address of the client as a new client.

3. The cluster control system according to claim 2, wherein

each of the plurality of slave computers further comprises a notification flag for determining whether or not to notify the IP address and the MAC address of the client;

and

the ARP notification unit determines whether or not to notify the IP address and the MAC address of the client by checking the notification flag.

4. The cluster control system according to claim 2, wherein

the master computer further comprises:

a client assignment table that stores assignment of a client to each of the plurality of slave computers; and

a slave node table that manages a status of each of the plurality of slave computers; and

the responsible node determination processing unit refers to the client assignment table and the slave node table when detecting that each of the plurality of slave computers has received an ARP request from the new client, adds the IP address and the MAC address of the new client to a responsible client table of one of the plurality of slave computers when the one of the plurality of slave computers is already assigned to the new client, and refers to the slave node table when none of the plurality of slave computers is assigned to the new client, selects one of the plurality of slave computers, and adds the IP address and the MAC address of the new client to a responsible client table of the selected slave computer.

5. The cluster control system according to claim 4, wherein

the responsible node determination processing unit adds or deletes an IP address and a MAC address of a client in the responsible client table of each of the plurality of slave computers to perform reassignment so that loads on the plurality of slave computers become equal, when detecting increase or decrease in the number of slave computers in the slave node table or when detecting unequal loads on the plurality of slave computers.

6. The cluster control system according claim 1, wherein

the responsible node determination processing unit transmits data including a command, an IP address, and a MAC address to each of the plurality of slave computers;

the command is classified into types of an addition command, a deletion command, and a clear command; and

each of the plurality of slave computers adds, deletes, or clears the IP address and the MAC address in the responsible client table according to the command when receiving the data including the command, the IP address, and the MAC address, sends an ARP response to a client specified by the IP address and the MAC address when the command is an addition command, and returns to the master computer information indicating whether or not the process according to the command has been normally finished.

7. A computer used as one of the plurality of slave computers in the cluster control system according to claim 1.

8. A computer used as the master computer in the cluster control system according to claim 1.

9. A cluster control method comprising:

connecting each of a plurality of clients and a server group through a front network;

connecting through a back-end network the server group with a master computer that controls the server group;

providing the front network with a single IP (Internet Protocol) address to cause a plurality of slave computers included in the server group to seem to have a same IP address;

by each of the plurality of slave computers, referring to a responsible client table indicating a client for which each of the plurality of slave computers is responsible when an ARP (Address Resolution Protocol) request for the single IP address has been received by the server group from each of the plurality of clients, sending an ARP response when a client that has sent the ARP request is present in the responsible client table, and not sending an ARP response when a client that has sent the ARP request is not present in the responsible client table; and

by the master computer, maintaining and managing through the back-end network in the responsible client table information regarding addition, change, or deletion of a client for which each of the plurality of slave computers is responsible and sending an ARP response to a client when the client is added.

10. The cluster control method according to claim 9, further comprising:

by each of the plurality of slave computers, when receiving an ARP request from a client that is not present in the responsible client table, notifying to the master computer an IP address and a MAC (Media Access Control) address of the client as a new client.

11. The cluster control method according to claim 10, further comprising:

determining whether or not to notify an IP address and a MAC address of the client by checking a notification flag for determining whether or not to notify the IP address and the MAC address of the client.

12. The cluster control method according to claim 10, further comprising:

providing the master computer with a client assignment table that stores assignment of a client to each of the plurality of slave computers;

providing the master computer with a slave node table that manages a status of each of the plurality of slave computers;

referring to the client assignment table and the slave node table when detecting that each of the plurality of slave computers has received an ARP request from the new client;

by the master computer, adding the IP address and the MAC address of the new client to a responsible client table of one of the plurality of slave computers when the one of the plurality of slave computers is already assigned to the new client; and

by the master computer, referring to the slave node table when none of the plurality of slave computers is assigned to the new client, selecting one of the plurality of slave computers, and adding the IP address and the MAC address of the new client to a responsible client table of the selected slave computer.

13. The cluster control method according to claim 12, further comprising:

adding or deleting an IP address and a MAC address of a client in a responsible client table of each of the plurality of slave computers to perform reassignment so that loads on the plurality of slave computers become equal, when detecting increase or decrease in the number of slave computers in the slave node table or when detecting unequal loads on the plurality of slave computers.

14. The cluster control method according to claim 9, further comprising:

by the master computer, transmitting data including a command, an IP address, and a MAC address to each of the plurality of slave computers; and

by each of the plurality of slave computers, adding, deleting, or clearing the IP address and the MAC address in the responsible client table according to the command when receiving the data including the command, the IP address, and the MAC address, and returning to the master computer information indicating whether or not the process according to the command has been normally finished, the command being classified into types of an addition command, a deletion command, and a clear command.

15. A program for causing a computer to execute an operation of each of the plurality of slave computers in the cluster control method according to claim 9.

16. A program for causing a computer to execute an operation of the master computer in the cluster control method according to claim 9.

17. A slave computer out of a plurality of slave computers that are connected to a plurality of clients through a front network and provide the front network with a single IP (Internet Protocol) address that is a same IP address, the slave computer comprising:

a responsible client table that indicates a client for which each of the plurality of slave computers is responsible; and

an ARP processing unit that refers to the responsible ;

30 client table when an ARP (Address Resolution Protocol) request for the single IP address has been received from one of the plurality of clients and sends an ARP response when a client that has sent the ARP request is present in the responsible client table, and otherwise does not send the ARP response, wherein

a client in the responsible client table is added, changed, or deleted by a master computer connected through a back-end network.

18. The slave computer according to claim 17, further comprising:

an ARP notification unit that, when receiving an ARP request from a client that is not present in the responsible client table, notifies to the master computer an IP address and a MAC (Media Access Control) address of the client as a new client.

19. The slave computer according to claim 18, further comprising a notification flag for determining whether or not to notify an IP address and a MAC address of the client, wherein

the ARP notification unit determines whether or not to notify the IP address and the MAC address of the client by checking the notification flag.

21. A cluster control system comprising:

the slave computer according to claim 17; and

a master computer connected through a back-end network to the plurality of slave computers, and controlling the plurality of slave computers, comprising a responsible node determination processing unit that performs through the back-end network addition, change, or deletion of a client in a responsible client table indicating a client for which each of the plurality of slave computers is responsible.

22. The cluster control system according to claim 21, wherein

the master computer further comprises:

a client assignment table that stores assignment of a client to each of the plurality of slave computers; and

a slave node table that manages a status of each of the plurality of slave computers; and

the responsible node determination processing unit refers to the client assignment table and the slave node table when detecting that each of the plurality of slave computers has received an ARP request from the new client, adds the IP address and the MAC address of the new client to a responsible client table of one of the plurality of slave computers when the one of the plurality of slave computers is already assigned to the new client, and refers to the slave node table when none of the plurality of slave computers is assigned to the new client, selects one of the plurality of slave computers, and adds the IP address and the MAC address of the new client to a responsible client table of the selected slave computer.

23. The cluster control system according to claim 22, wherein

the responsible node determination processing unit adds or deletes an IP address and a MAC address of a client in a responsible client table of each of the plurality of slave computers to perform reassignment so that loads on the plurality of slave computers are equal, when detecting increase or decrease in the number of slave computers in the slave node table or when detecting unequal loads on the plurality of slave computers.

24. The cluster control system according to claims 21, wherein

the responsible node determination processing unit transmits data including a command, an IP address, and a MAC address to each of plurality of the slave computers;

the command is classified into types of an addition command, a deletion command, and a clear command; and

each of the plurality of slave computers adds, deletes, or clears the IP address and the MAC address in the responsible client table according to the command when receiving the data including the command, the IP address, and the MAC address, and transmits an ARP response to a client specified by the IP address and the MAC address when the command is an addition command, and returns to the master computer information indicating whether or not the process according to the command has been normally finished.

20. A master computer connected through a back-end network to a plurality of slave computers that are connected to a plurality of clients through a front network and provide the front network with a single IP address that is a same IP (Internet Protocol) address, and controlling the plurality of slave computers, the master computer comprising:

a responsible node determination processing unit that performs through the back-end network addition, change, or deletion of a client in a responsible client table indicating a client for which each of the plurality of slave computers is responsible.

25. A cluster control method comprising:

by each of a plurality of slave computers that are connected to a plurality of clients through a front network and provide the front network with a single IP (Internet Protocol) address that is a same IP address, referring to a responsible client table indicating a client for which each of the plurality of slave computers is responsible when an ARP (Address Resolution Protocol) request for the single IP address has been received from one of the plurality of clients, sending an ARP response when a client that has sent the ARP request is present in the responsible client table, and otherwise not sending an ARP response; and

having a client in the responsible client table added, changed, or deleted by a master computer connected through a back-end network.

26. A program causing each of a plurality of slave computers that are connected to a plurality of clients through a front network and provide the front network with a single IP (Internet Protocol) address that is a same IP address, to execute:

referring to a responsible client table indicating a client for which each of the plurality of slave computers is responsible when an ARP (Address Resolution Protocol) request for the single IP address has been received from one of the plurality of clients, sending an ARP response when a client that has sent the ARP request is present in the responsible client table, and otherwise not sending an ARP response; and

having a client in the responsible client table added, changed, or deleted by a master computer connected through a back-end network.