Methods, systems and computer program products for workload distribution based on end-to-end quality of service
First Claim
1. A method of distributing workload between data processing systems executing an application which communicates over a network, the method comprising:
- receiving a request for a connection to the application over the network;
obtaining workload information for the data processing systems;
obtaining network quality of service information associated with communications over the network for respective ones of the data processing systems;
generating workload metrics associated with respective ones of the data processing systems utilizing the workload information and the corresponding network quality of service information for the data processing systems;
distributing the requested connection to instances of the application executing on the data processing systems based on the generated workload metrics;
wherein the workload information comprises a weight value (W) corresponding to a data processing systems processing capacity;
wherein the network quality of service information comprises a loss weight (F-loss) based on packet loss ratio, a network timeout weight (F-timeout) based on a network timeout ratio and a connection weight (F-con) based on a number of active connections;
determining a value for F-loss by mapping a TCP loss ratio to the value for F-loss;
determining a value for F-timeout by mapping a timeout loss ratio to the value for F-timeout;
determining a value for F-conn by determining if a number of current connections exceed a predefined percentage of a total number of allowed connections and setting F-con to 1 if the total number of connections exceeds the predefined percentage of the total number of allowed connections of the total number of allowed connections; and
wherein generating workload metrics comprises evaluating the equation;
Aw=(1−
MIN(1,F-loss+timeout+F-con)*W where Aw is the workload metric.
1 Assignment
0 Petitions
Accused Products
Abstract
Methods, systems and computer program products provide for distributing workload between data processing systems executing an application which communicates over a network, by receiving a request for a connection to the application over the network, obtaining workload information for the data processing systems, obtaining network quality of service (QoS) information associated with communications over the network for the data processing systems and utilizing the workload information and the corresponding network quality of service information for the data processing systems so as to provide workload metrics. The requested connection to the application is distributed based on the generated workload metrics.
-
Citations
18 Claims
-
1. A method of distributing workload between data processing systems executing an application which communicates over a network, the method comprising:
-
receiving a request for a connection to the application over the network; obtaining workload information for the data processing systems; obtaining network quality of service information associated with communications over the network for respective ones of the data processing systems; generating workload metrics associated with respective ones of the data processing systems utilizing the workload information and the corresponding network quality of service information for the data processing systems; distributing the requested connection to instances of the application executing on the data processing systems based on the generated workload metrics; wherein the workload information comprises a weight value (W) corresponding to a data processing systems processing capacity; wherein the network quality of service information comprises a loss weight (F-loss) based on packet loss ratio, a network timeout weight (F-timeout) based on a network timeout ratio and a connection weight (F-con) based on a number of active connections; determining a value for F-loss by mapping a TCP loss ratio to the value for F-loss; determining a value for F-timeout by mapping a timeout loss ratio to the value for F-timeout; determining a value for F-conn by determining if a number of current connections exceed a predefined percentage of a total number of allowed connections and setting F-con to 1 if the total number of connections exceeds the predefined percentage of the total number of allowed connections of the total number of allowed connections; and wherein generating workload metrics comprises evaluating the equation;
Aw=(1−
MIN(1,F-loss+timeout+F-con)*Wwhere Aw is the workload metric. - View Dependent Claims (2, 3, 4, 5, 6)
-
-
7. A system for workload distribution, comprising:
-
a workload distributer which selects data processing systems in a cluster of data processing systems for distribution of connections based on network quality of service information associated with the data processing systems and workload information associated with the data processing systems; a router operably associated with the workload distributer which receives requests for connection to an application executing on ones of the data processing systems and distributes the connections to data processing systems in the cluster of data processing systems selected by the workload distributer; wherein the network quality of service information comprises a loss weight (F-loss) based on packet loss ratio, a network timeout weight (F-timeout) based on a network timeout ratio and a connection weight (F-con) based on a number of active connections; wherein F-loss is determined by mapping a TCP loss ratio to the value for F-loss, F-timeout is determined by mapping a timeout loss ratio to the value for F-timeout and F-conn has a value of 1 if a total number of connections to the application associated with the connection request exceeds a predefined percentage of the total number of allowed connections and 0 if the total number of connections does not exceed the predefined percentage of the total number of allowed connections; wherein the workload distributer is configured to evaluate the equation;
Aw=(1−
MIN(1,F-loss+F-timeout+F-con)*Wwhere Aw is a workload metric and W is the workload information; and wherein the workload distributor selects a data processing system with a best Aw value. - View Dependent Claims (8, 9, 10)
-
-
11. A system for distributing workload between data processing systems executing an application which communicates over a network, comprising:
-
means for receiving a request for a connection to the application over the network; means for obtaining workload information for the data processing systems; means for obtaining network quality of service information associated with communications over the network for respective ones of the data processing systems; means for generating workload metrics associated with respective ones of the data processing systems utilizing the workload information and the corresponding network quality of service information for the data processing systems; means for distributing the requested connection to instances of the application executing on the data processing systems based on the generated workload metrics; wherein the workload information comprises a weight value (W) corresponding to a data processing systems processing capacity; wherein the network quality of service information comprises a loss weight (F-loss) based on packet loss ratio, a network timeout weight (F-timeout) based on a network timeout ratio and a connection weight (F-con) based on a number of active connections; means for determining a value for F-loss by mapping a TCP loss ratio to the value for F-loss; means for determining a value for F-timeout by mapping a timeout loss ratio to the value for F-timeout; means for determining a value for F-conn by determining if a number of current connections exceed a predefined percentage of a total number of allowed connections and setting F-con to 1 if the total number of connections exceeds the predefined percentage of the total number of allowed connections and to 0 if the total number of connections does not exceed the predefined percentage of the total number of allowed connections; and wherein the means for generating workload metrics comprises means for evaluating the equation;
Aw=(1−
MIN(1,F-loss+F-timeout+F-con)*Wwhere Aw is the workload metric. - View Dependent Claims (12, 13, 14)
-
-
15. A computer program product for distributing workload between data processing systems executing an application which communicates over a network, comprising:
-
a computer readable storage medium having computer readable program code embodied therein, the computer readable program code comprising; computer readable program code which receives a request for a connection to the application over the network; computer readable program code which obtains workload information for the data processing systems; computer readable program code which obtains network quality of service information associated with communications over the network for respective ones of the data processing systems; computer readable program code which generates workload metrics associated with respective ones of the data processing systems utilizing the workload information and the corresponding network quality of service information for the data processing systems; computer readable program code which distributes the requested connection to instances of the application executing on the data processing systems based on the generated workload metrics; wherein the workload information comprises a weight value (W) corresponding to a data processing systems processing capacity; wherein the network quality of service information comprises a loss weight (F-loss) based on packet loss ratio, a network timeout weight (F-timeout) based on a network timeout ratio and a connection weight (F-con) based on a number of active connections; computer readable program code which determines a value for F-loss by mapping a TCP loss ratio to the value for F-loss; computer readable program code which determines a value for F-timeout by mapping a timeout loss ratio to the value for F-timeout; and computer readable program code which determines a value for F-conn by determining if a number of current connections exceed a predefined percentage of a total number of allowed connections and setting F-con to 1 if the total number of connections exceeds the predefined percentage of the total number of allowed connections and to 0 if the total number of connections does not exceed the predefined percentage of the total number of allowed connections; and wherein the computer readable program code which generates workload metrics comprises computer readable program code which evaluates the equation;
Aw=(1−
MIN(1, F-loss+F-timeout+F-con))*Wwhere Aw is the workload metric. - View Dependent Claims (16, 17, 18)
-
Specification