Load Balancer and Load Balancing System

• US 20110276982A1
• Filed: 03/01/2011
• Published: 11/10/2011
• Est. Priority Date: 05/06/2010
• Status: Active Grant

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: Nakayama, Mariko, Toumura, Kunihiko, Kumagai, Atsuya, Sugauchi, Kiminori
• Granted Patent: US 8,656,406 B2
• Time in Patent Office: Days
• US Class Current: 718/105
• CPC Class Codes: G06F 9/505 considering the load
• 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

Abstract

In a system including a load balancer to select a virtual server to which a request is to be transferred, the load balancer includes a function to monitor resource use states of physical and virtual servers and a function to predict a packet loss occurring in a virtual switch. The request count of requests processible by each virtual server is calculated based on the resource amount available for the virtual server and a packet loss rate of the virtual switch, to thereby select a virtual server capable of processing a larger number of requests.

Claims

1. A load balancer disposed for one physical server or a plurality of physical servers in which a plurality of virtual servers operate under control of a virtual operating system disposed in the physical server or servers, the load balancer distributing requests to a plurality of virtual servers under control thereof, comprising:

a resource monitoring unit for monitoring, for each of the physical servers, a CPU utilization of each of the virtual servers and a CPU utilization of the physical server;

a resource management table for keeping therein, for each of the virtual servers, at least a CPU assignment rate set to the virtual server and designation of a second virtual server which shares a resource with the virtual server; and

a server selecting unit for calculating, in the presence of a request to be transferred, for each transfer destination candidate virtual server, an amount of the CPU resource available for the virtual server based on a total of the CPU utilization of the virtual server and the CPU utilization of the second virtual server designated to share the resource with the virtual server and a total of the CPU assignment rate set to the virtual server and a CPU utilization set to the second virtual server, and selecting one of the virtual servers to which the request is to be transferred, based on the amount of the CPU resource available for each transfer destination candidate virtual server.

2. A load balancer according to claim 1, further comprising

a packet loss rate predicting unit formed in each of the physical servers for predicting a packet loss rate of a virtual switch connecting each virtual server in the physical server to an external unit, wherein

the server selecting unit comprises a function for changing a scheme of selecting a virtual server in association with the packet loss rate predicted by the packet loss rate predicting unit.

3. A load balancer according to claim 1, further comprising

a packet loss rate predicting unit formed in each of the physical servers for predicting a packet loss rate of a virtual switch connecting each virtual server in the physical server to an external unit, wherein

the server selecting unit multiplies the available amount of the CPU resource predicted for each of the transfer destination candidate virtual servers by a reciprocal of the packet loss rate predicted for the virtual switch connecting to the transfer destination candidate virtual server, to thereby calculate processing capacity of each transfer destination candidate virtual server, and selects, based on the processing capacity thus calculated, one of the virtual servers to which the request is to be transferred.

4. A load balancer according to claim 2, wherein the packet loss rate predicting unit calculates the packet loss rate based on the CPU utilization of the physical server and an input packet rate of packets to be inputted to the physical server.

5. A load balancer according to claim 2, wherein the packet loss rate predicting unit keeps therein a correspondence between an input packet rate of packets to be inputted to the physical server, the CPU utilization of the physical server, and the packet loss rate occurring in the virtual switch, to thereby calculate the packet loss rate based on the correspondence.

6. A load balancer according to claim 2, wherein the packet loss rate predicting unit keeps therein a correspondence between an input packet rate of packets to be inputted to the physical server, the CPU utilization of the physical server, and the packet loss rate occurring in the virtual switch, and multiplies the packet loss rate obtained from the correspondence by a correction value associated with a current CPU utilization of the physical server, to thereby calculate the packet loss rate of the virtual switch.

7. A load balancer according to claim 2, wherein the server selecting unit calculates, based on the CPU assignment rate of the virtual server and the packet loss rate of the virtual switch calculated by the packet loss rate predicting unit, a request count of requests processible by the virtual server, and compares with each other the request counts thus calculated for the respective virtual servers of request transfer destination candidates, to thereby select one of the virtual servers to which the request is to be transferred.

8. A load balancer according to claim 1, wherein the server selecting unit calculates, based on the CPU assignment rate of the virtual server and the CPU utilization of the physical server, a request count of requests processible by the virtual server, and compares with each other the request counts thus calculated for the respective virtual servers of request transfer destination candidates, to thereby select one of the virtual servers to which the request is to be transferred.

9. A load balancer disposed for one physical server or a plurality of physical servers in which a plurality of virtual servers operate under control of a virtual operating system disposed in the server or servers, the load balancer distributing requests to a plurality of virtual servers under control thereof, comprising:

a resource monitoring unit for monitoring, for each of the physical servers, a CPU utilization of each of the virtual servers and a CPU utilization of the physical server;

a resource management table for keeping therein, for each of the virtual servers, at least a CPU assignment rate set to the virtual server; and

a request count conversion table for keeping therein, for each of the virtual servers, a value of a request processing count for the virtual server corresponding to each of combinations of a plurality of values indicating the CPU utilization of the physical server forming the virtual server and a plurality of values indicating the CPU assignment rate of the virtual server, wherein

in the presence of a request to be transferred, the load balancer, for each request transfer destination candidate virtual server, searches the request count conversion table based on a value of the CPU utilization of the physical server forming the virtual server obtained from the resource monitoring unit and a value of the CPU assignment rate of the virtual server obtained from the resource management table, to calculate a request count of requests currently processible by the virtual server, and compares with each other the request counts of processible requests thus calculated for the respective virtual servers, to thereby select one of the virtual servers to which the request is to be transferred.

10. A load balancing method for use in a request processing system comprising one physical server or a plurality of physical servers in which a plurality of virtual servers operate under control of a virtual operating system disposed in the physical server or servers and a load balancer disposed for distributing requests to a plurality of virtual servers operating in the physical servers, comprising the steps of:

beforehand keeping, for each of the virtual servers, setting of a CPU assignment rate and designation of a second virtual server which shares a resource with the virtual server;

monitoring, for each of the physical servers, a CPU utilization of each of the virtual servers and a CPU utilization of the physical server; and

in the presence of a request to be transferred, calculating, for each transfer destination candidates virtual server, an amount of the CPU resource available for the virtual server based on a total of the CPU utilization of the virtual server and the CPU utilization of the second virtual server and a total of the CPU assignment rate of the virtual server and a CPU utilization of the second virtual server, and selecting one of the virtual servers to which the request is to be transferred, based on the amount of the CPU resource available for the virtual server calculated.