Load balancing using distributed forwarding agents with application based feedback for different virtual machines

• US 6,970,913 B1
• Filed: 07/02/1999
• Issued: 11/29/2005
• Est. Priority Date: 07/02/1999
• Status: Expired due to Term

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Inventors: Albert, Mark, Tiwari, Pranav Kumar, O'Rourke, Chris, Menditto, Louis F., Howes, Richard A., Tsang, Tzu-Ming, Kersey, Edward A.
• Primary Examiner(s): Harvey, Jack B.
• Assistant Examiner(s): Nguyen, Hai V.
• Application Number:
  US09/347,111
• Time in Patent Office: 2,342 Days
• Field of Search: 709/217, 709/218, 709/219, 709/200, 709/223, 709/227, 709/228, 340/825.25, 375/219, 711/1, 370/389, 370/392, 370/449
• US Class Current: 709/217
• 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

Abstract

A system and method are described for selecting a server to handle a connection. The method includes receiving at a service manager a connection request intercepted by a network device having a forwarding agent that is operative to receive instructions from a service manager, the connection request having been forwarded from the forwarding agent on the network device to the service manager. A preferred server is selected at the service manager from among a group of available servers. The preferred server is the server that is to service the connection request. Instructions are sent from the service manager to the forwarding agent. The instructions include the preferred server that is to service the connection request so that the connection request may be forwarded from the network device to the preferred server.

Claims

1. A method of selecting a server to handle a connection, comprising:

receiving at a service manager at a first network node a connection request intercepted by a particular one of a plurality of forwarding agents that are each configured to intercept a subset of the connection requests associated with a network address and operative to receive instructions from the service manager, at least one of the plurality of forwarding agents being located at a second network node different than the first network node, the plurality of forwarding agents being configured to intercept collectively all connection requests associated with the network address by each being configured to intercept a subset of said connection requests, such that the task of intercepting and processing said connection requests is distributed among them, the connection request having been forwarded from the particular one of the plurality of forwarding agents to the service manager;

selecting at the service manager a preferred server from among a group of available servers, the preferred server being the server that is to service the connection request;

sending instructions from the service manager to the particular one of the plurality of forwarding agents, the instructions including the preferred server that is to service the connection request so that the connection request may be forwarded from the particular one of the plurality of forwarding agents to the preferred server, the instructions operable to allow the particular one of the forwarding agents to forward further requests associated with the connection request without further intervention by the service manager;

whereby connection requests associated with the network address may be processed in a distributed manner, without requiring that all connection requests associated with the network address be processed by a single appliance forwarding agent located at a single node through which all such connection requests must pass.

2. A method of selecting a server to handle a connection as recited in claim 1 wherein the particular one of the plurality of forwarding agents is a router.

3. A method of selecting a server to handle a connection as recited in claim 1 wherein the particular one of the plurality of forwarding agents is a switch.

4. A method of selecting a server to handle a connection as recited in claim 1 wherein sending instructions from the service manager to the particular one of the plurality of forwarding agents further includes sending the connection request back to the particular one of the plurality of forwarding agents to be forwarded to the preferred server.

5. A method of selecting a server to handle a connection as recited in claim 1 wherein the connection request is a TCP SYN packet.

6. A method of selecting a server to handle a connection as recited in claim 1 wherein a plurality of individual servers in the group of available servers determine respective capacity weights, the capacity weights indicating a capacity of each individual server to process connections.

7. A method of selecting a server to handle a connection as recited in claim 6 wherein the plurality of individual servers in the group of available servers includes all of the servers in the group of available servers.

8. A method of selecting a server to handle a connection as recited in claim 6 wherein the plurality of individual servers in the group of available servers includes a subset of the servers in the group of available servers.

9. A method of selecting a server to handle a connection as recited in claim 6 wherein the plurality of individual servers send the capacity weights to the service manager.

10. A method of selecting a server to handle a connection as recited in claim 6 wherein the plurality of individual servers send the capacity weights to a selected one of the plurality of individual servers and the selected one of the plurality of individual servers sends the capacity weights to the service manager.

11. A method of selecting a server to handle a connection as recited in claim 6 wherein the capacity weights are normalized weights that indicate the capacity of each individual server to process connections relative to other servers of the plurality of individual servers.

12. A method of selecting a server to handle a connection as recited in claim 6 wherein the capacity weights include different weights for different ports on one of the plurality of individual servers.

13. A method of selecting a server to handle a connection as recited in claim 6 wherein one of the plurality of individual servers services a plurality of virtual IF addresses and wherein the capacity weights include different weights for the one of the plurality of individual servers for each of the plurality of virtual IF addresses.

14. A method of selecting a server to handle a connection as recited in claim 6 wherein capacity the weights are sent to the service manager.

15. A method of selecting a server to handle a connection as recited in claim 6 wherein selecting at the service manager a preferred server from among a group of available servers includes selecting a server from among the group of available servers in a manner such that the capacity weights influence the selection of a server from among the plurality of individual servers.

16. A method of selecting a server to handle a connection as recited in claim 6 wherein selecting at the service manager a preferred server from among a group of available servers includes randomly selecting a server from among the group of available servers in a manner such that the capacity weights determine the probability of selecting a server from among the plurality of individual servers.

17. A method of selecting a server to handle a connection as recited in claim 6 wherein selecting at the service manager a preferred server from among a group of available servers includes selecting a server from among the group of available servers by selecting the server currently handling a least weighted number of connections wherein the weighted number of connections is determined by applying a weighting factor derived from one of the capacity weights to the number of connections handled by each server in the plurality of servers.

18. A method of selecting a server to handle a connection as recited in claim 6 wherein the capacity weights are periodically sent from the plurality of individual servers to the service manager.

19. A method of selecting a server to handle a connection as recited in claim 6 wherein the capacity weights are sent from an individual server when an event occurs on the individual server.

20. A method of selecting a server to handle a connection as recited in claim 19 wherein the event is a spike in utilization of the individual server.

21. A method of selecting a server to handle a connection as recited in claim 19 wherein the event is a loss of a port of the individual server.

22. A method of selecting a server to handle a connection from among a group of available servers including:

receiving at a first network node a message that includes a plurality of weights, each of the plurality of weights indicating a capacity of an individual server in the group of available servers to process connections corresponding to a plurality of virtual IP addresses;

determining a preferred server from among the group of available servers for handling an incoming connection based on the plurality of weights; and

directing a a particular one of a plurality of forwarding agents upon initiation of the connection to forward packets for the connection to the preferred server;

forwarding packets from the particular one of the plurality of forwarding agents to the preferred server without further intervention external to the particular one of the plurality of forwarding agents;

wherein the plurality of forwarding agents are configured to intercept collectively all connection requests associated with at least one virtual IP address of the plurality of virtual IP addresses by each being configured to intercept a subset of said connection requests, such that the task of intercepting and processing said connection requests is distributed among them.

23. A method of selecting a server to handle a connection from among a group of available servers as recited in claim 22 wherein the plurality of weights indicate the capacity of an individual server in the group of available servers to process connections corresponding to a plurality of applications.

24. A method of selecting a server to handle a connection from among a group of available servers as recited in claim 23 wherein the plurality of applications correspond to a plurality of ports.

25. A method of selecting a server to handle a connection from among a group of available servers as recited in claim 23 wherein the plurality of applications correspond to a plurality of protocols.

26. A service manager that selects a server to handle a connection from among a group of available servers including:

a first communications interface configured to receive at a first network node and process messages that include a plurality of weights, the plurality of weights indicating the capacity of an individual server in the group of available servers to process connections corresponding to a plurality of virtual IP addresses;

a database that is configured to store the weights from the messages;

a processor that is configured to select a preferred server from among the group of available servers for handling an incoming connection based on the stored weights; and

a second communications interface configured to communicate the preferred server to a forwarding agent, the forwarding agent operable to forward packets to the preferred server without further intervention by the service manager;

wherein the forwarding agent comprises one of a plurality of forwarding agents, at least one of which is located at a second network node different than the first network node, the plurality of forwarding agents being configured to intercept collectively all connection requests associated with at least one virtual IP address of the plurality of virtual IP addresses by each being configured to intercept a subset of said connection requests, such that the task of intercepting and processing said connection requests is distributed among them.

27. A service manager that selects a server to handle a connection from among a group of available servers as recited in claim 26 wherein the first communications interface and the second communications interface are included in a single combined interface.

28. A computer program product for selecting a server to handle a connection embodied in a computer readable medium comprising computer instructions for:

receiving a connection request at a forwarding agent that is configured to intercept a subset of the connection requests associated with a network address and operative to receive instructions from a service manager located at a first network node;

forwarding the connection request from the forwarding agent to the service manager;

selecting at the service manager a preferred server from among a group of available servers, the preferred server being the server that is to service the connection request;

sending instructions from the service manager to the forwarding agent, the instructions including the preferred server that is to service the connection request; and

forwarding the connection request from the forwarding agent to the preferred server;

forwarding further requests associated with the connection request from the forwarding agent to the preferred server without further intervention by the service manager;

wherein the forwarding agent comprises one of a plurality of forwarding agents, at least one of which is located at a second network node different than the first network node, the plurality of forwarding agents being configured to intercept collectively all connection requests associated with the network address by each being configured to intercept a subset of said connection requests, such that the task of intercepting and processing said connection requests is distributed among them.

29. A computer program product for selecting a server to handle a connection from among a group of available servers embodied in a computer readable medium comprising computer instructions for:

receiving at a first network node messages that include a plurality of weights, the plurality of weights indicating a capacity of each individual server in the group of available servers to process connections corresponding to a plurality of virtual IP addresses;

storing the weights from the messages;

selecting a preferred server from among the group of available servers for handling an incoming connection based on the stored weights; and

communicating the preferred server to a forwarding agent;

forwarding packets from the forwarding agent to the preferred server without further intervention external to the forwarding agent;

wherein the forwarding agent comprises one of a plurality of forwarding agents, at least one of which is located at a second network node different than the first network node, the plurality of forwarding agents being configured to intercept collectively all connection requests associated with at least one virtual IP address of the plurality of virtual IP addresses by each being configured to intercept a subset of said connection requests, such that the task of intercepting and processing said connection requests is distributed among them.