On-demand global server load balancing system and method of use

• US 8,539,075 B2
• Filed: 08/11/2006
• Issued: 09/17/2013
• Est. Priority Date: 04/21/2006
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Bali, Bahri B, Greenlee, Gordan G, Weingarten, Richard E
• Primary Examiner(s): Nawaz, Asad
• Assistant Examiner(s): Ansari, Najeebuddin
• Application Number:
  US11/464,097
• Publication Number:
  US 20070250631A1
• Time in Patent Office: 2,594 Days
• Field of Search: 709223-226
• US Class Current: 709/226
• CPC Class Codes: G06F 11/3433 for load management allocation of a server based on load conditions G06F9/505; load rebalancing G06F9/5083; redistributing the load in a network by a load balancer H04L67/1029
• CPC Class Codes: H04L 41/5019 Ensuring fulfilment of SLA
• CPC Class Codes: H04L 43/0817 by checking functioning
• CPC Class Codes: H04L 43/0876 Network utilisation, e.g. volume of load or congestion level
• CPC Class Codes: H04L 43/16 Threshold monitoring
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• 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 67/1021 based on client or server locations
• CPC Class Codes: H04L 67/1023 based on a hash applied to IP addresses or costs

Abstract

A method and system for measuring performance and balancing workload of sites and servers is provided. The method comprises measuring performance of one or more servers located at different sites of the network and computing a performance metric for each of the sites or one or more servers at each of the sites. The method further comprises providing a bias to the performance metric and routing a request to one of the different sites or one of the one or more servers based on the biased performance metric.

Claims

1. A method of processing requests in a network, comprising:

measuring performance of one or more servers located at different sites of the network;

computing a load balancing performance metric for each of the sites or the one or more servers at each of the sites;

providing a bias to the load balancing performance metric; and

routing a request to one of the different sites or one of the one or more servers based on the biased load balancing performance metric to balance a workload among the one or more servers,

wherein the bias is a multiplier that accentuates a load issue by proactively taking into consideration overall architecture of the network in order to adjust the workload among the one or more servers prior to any capacity problems;

the load balancing performance metric is a load value; and

the load value is calculated by;

factor=(upperbound−

lowerbound)/critical response time; and

load=round((factor*response time)+lowerbound).

2. The method of claim 1, further comprising computing a proximity metric based on a distance between each of the different sites and a requestor.

3. The method of claim 1, wherein the load balancing performance metric includes the response time from the one or more servers at the sites.

4. The method of claim 1, wherein the biased load balancing performance metric is weighted.

5. The method of claim 1, further comprising:

routing the request to a site selector for selecting a site from the different sites;

selecting the site for processing the request based on the biased load balancing performance metric; and

routing the request to an address associated with the site for processing.

6. The method of claim 5, wherein the address is an address of a network dispatcher.

7. The method of claim 6, wherein the routing comprises:

selecting one server of a plurality of servers at the site; and

routing the request to the one server.

8. The method of claim 1, further comprising storing the site load balancing performance metric on at least one site selector.

9. The method of claim 1, wherein the bias is based on an architecture of the network.

10. The method of claim 9, wherein the architecture includes at least one of networking speed, number of servers at each site, types of servers and their specific configurations.

11. The method of claim 1, wherein the load value is biased to:

proactively identify sites of the different sites that have a capacity to respond to The request; and

automatically direct traffic away from sites which are running close to, at or over capacity.

12. The method of claim 11, wherein the load balancing performance metric is computed based on response times that are greater than the critical response time, which is a threshold used to determine whether the different sites or the one or more servers need to be marked down or not.

13. The method of claim 12, further comprising providing a limit on how many consecutive response times that are greater than the critical response time can be set, wherein:

until the limit is exceeded, a last valid biased load value is used to route traffic;

if a response time value cannot be obtained, a last biased load value cached from a previous cycle is used to route traffic; and

a default biased load value is used to route traffic.

14. The method of claim 13, further compromising time stamping a last two response time values and making a comparison thereof, wherein if the comparison provides a same value, the last biased load value is used to route traffic.

15. The method of claim 1, wherein the steps of claim 1 are provided on a subscription, advertising, and/or fee basis.

16. The method of claim 1, wherein the bias multiplier is highest for a server of the one or more servers that has a highest capacity with relation to other servers of the one or more servers.

36. The method of claim 1, wherein providing a bias to the load balancing performance metric comprises adjusting the load balancing performance metric by the bias multiplier to determine the biased load balancing performance metric.

37. The method of claim 36, wherein the bias multiplier is user-supplied.

38. The method of claim 37, wherein the bias multiplier accounts for networking speed, an amount of servers at a respective site, types of servers at the respective site and overall configuration of servers at the respective site.

17. A method of routing requests in a network, comprising:

computing a load balancing performance metric for at least one of a site and a server of a plurality of servers; and

routing a request to at least the site or the server based on the load balancing performance metric for balancing workload among the plurality of servers,

wherein the load balancing performance metric includes a biased response time metric associated with the site or the server of the plurality of servers,

wherein the biased response time metric comprises a computed response time changed by a bias factor, and

wherein the bias factor is a multiplier that accentuates a load issue by proactively taking into consideration overall architecture of the network in order to adjust the workload among the one or more servers prior to any capacity problems;

the load balancing performance metric is a load value; and

the load value is calculated by;

the bias factor=(upperbound−

lowerbound)/critical response time and

load=round((the bias factor*the response time)+lowerbound).

18. A system implemented on hardware for routing requests in a network, comprising:

a computer infrastructure configured to;

measure performance of one or more servers at different sites;

generate a load balancing performance metric for each of the one or more servers;

provide a bias to the load balancing performance metric by altering the load balancing performance metric by a bias factor to generate a biased load balancing performance metric; and

route a request to one of the different sites or one of the one or more servers based on the biased load balancing performance metric to balance a workload among the one or more servers,

wherein the bias factor is a multiplier that accentuates a load issue by proactively taking into consideration overall architecture of the network in order to adjust the workload among the one or more servers prior to any capacity problems;

the load balancing performance metric is a load value; and

the load value is calculated by;

the bias factor=(upperbound−

lowerbound)/critical response time; and

load=round((the bias factor*response time)+lowerbound); and

the providing the bias is implemented using at least one processor of the computer infrastructure.

19. The system of claim 18, wherein:

the computer infrastructure is further configured to store stores the load balancing performance metric;

the computer infrastructure comprises an enterprise directory resource monitor (ERDM) that monitors the system to ascertain relative locations of requests, measures individual response time, and computes average site response time; and

the ERDM comprises a probe daemon, a console daemon, and a status daemon.

20. The system of claim 18, wherein the computer infrastructure is further configured to compute a proximity metric based on a distance between the different sites and a requestor.

21. The system of claim 20, wherein the load balancing performance metric includes the response time from the one or more servers at the sites, which is weighted.

22. The system of claim 18, wherein the computer infrastructure is further configured to:

route the request to a site selector for selecting a site from the sites;

select the site for processing the request based on the biased load balancing performance metric; and

route the request to an address associated with the site for processing.

23. The system of claim 18, wherein the computer infrastructure is further configured to store the load balancing performance metric on at least one site selector.

24. The system of claim 18, wherein the bias is based on at least one of networking speed, number of servers at each site, types of servers and their specific configurations.

25. The system of claim 18, wherein the load value is biased to:

proactively identify those sites of the different sites that have a capacity to respond to the request; and

automatically direct traffic away from sites which are running close to, at or over capacity.

26. The system of claim 18, wherein the load balancing performance metric is computed based on response times that are greater than the critical response time, which is a threshold used to determine whether the different sites or the one or more servers need to be marked down or not.

27. The system of claim 26, wherein the computer infrastructure is further configured to provide a limit on how many consecutive response times that are greater than the critical response time can be set, wherein:

until the limit is exceeded, a last valid biased load value is used to route traffic;

if a response time value cannot be obtained, a last biased load value cached from a previous cycle is used to route traffic; and

a default biased load value is used to route traffic.

28. The system of claim 27, wherein the computer infrastructure is further configured to time stamp a last two response time values and make a comparison thereof, wherein if they are a same value, the last valid biased load value is used to route traffic.

29. A system for balancing workload, comprising:

a computing device comprising a combination of hardware and software configured to;

collect performance measurements for different sites in a network;

generate at least one biased server load balancing performance metric based on a response time of servers located at the different sites; and

route a request to a server at one of the different sites based on the at least one biased server load balancing performance metric,

wherein the at least one biased server load balancing performance metric comprises a respective response time of the servers located at the different sites changed by a bias factor, and

wherein the bias factor is a multiplier that accentuates a load issue by proactively taking into consideration overall architecture of the network in order to adjust workload among the one or more servers prior to any capacity problems;

the load balancing performance metric is a load value; and

the load value is calculated by;

the bias factor=(upperbound−

lowerbound)/critical response time; and

load=round((the bias factor*the response time)+lowerbound).

30. A method to provide a service for establishing a computing infrastructure, comprising:

collecting performance measurements for different sites in a network;

generating at least one biased server load balancing performance metric based on the performance measurements of servers located at the different sites; and

routing a request to a server at one of the different sites based on the biased server load balancing performance metric,

wherein the at least one biased server load balancing performance metric comprises a respective performance measurement of the servers located at the different sites changed by a bias factor, and

wherein the bias factor is a multiplier that accentuates a load issue by proactively taking into consideration overall architecture of the network in order to adjust workload among the one or more servers prior to any capacity problems;

the load balancing performance metric is a load value; and

the load value is calculated by;

the bias factor=(upperbound−

lowerbound)/critical response time; and

load=round((the bias factor*response time)+lowerbound).

31. The method of claim 30, wherein the performance measurements are load balancing measurements of the different sites using at least response times.

32. The method of claim 30, further comprising determining a proximity between a requestor and the different sites, and using the proximity with the biased server performance metric to route the request.

33. The method of claim 32, wherein the proximity and the server load balancing performance metric are weighted.

34. A method of processing queries in a network, comprising the steps of:

measuring performance of one or more servers at different locations;

routing a request to the one of the one or more servers based on a computed load balancing performance metric including proximity to the requestor and a biased allocation of workload amongst the one or more servers,

wherein the biased allocation of workload comprises a computed allocation of workload changed by a bias factor, and

wherein the bias is a multiplier that accentuates a load issue by proactively taking into consideration overall architecture of the network in order to adjust the workload among the one or more servers prior to any capacity problems;

the load balancing performance metric is a load value; and

the load value is calculated by;

factor=(upperbound−

lowerbound)/critical response time; and

load=round((factor*response time)+lowerbound).

35. A computer program product comprising a computer usable device having readable program code embodied in the device, the computer program product includes at least one processor to:

collect performance measurements for different sites in a network;

generate at least one biased server load balancing performance metric based on the performance measurements of servers located at the different sites; and

route a request to a server at one of the different sites based on the biased server load balancing performance metric,

wherein the at least one biased server load balancing performance metric comprises a respective performance measurement of the servers located at the different sites changed by a bias factor, and

wherein the bias factor is a multiplier that accentuates a load issue by proactively taking into consideration overall architecture of the network in order to adjust workload among the one or more servers prior to any capacity problems;

the load balancing performance metric is a load value; and

the load value is calculated by;

the bias factor=(upperbound−

lowerbound)/critical response time; and

load=round((the bias factor*response time)+lowerbound).