Generating automated mappings of service demands to server capacites in a distributed computer system

• US 20030236822A1
• Filed: 06/10/2002
• Published: 12/25/2003
• Est. Priority Date: 06/10/2002
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Graupner, Sven, Kotov, Vadim
• Granted Patent: US 7,072,960 B2
• Time in Patent Office: Days
• US Class Current: 709/203
• CPC Class Codes: G06F 9/5066 Algorithms for mapping a plurality of inter-dependent sub-tasks onto a plurality of physical CPUs mappping at compile time, see G06F8/451
• CPC Class Codes: G06F 9/5072 Grid computing

Abstract

A method for generating automated mappings of service demands to server capacities in a distributed computer system, including a plurality of server and service nodes. The method establishing server model layer and a service model layer and a set of server and service attribute values, which are associated with the plurality of server and service nodes. Each set of server attributes values and service attributes values are normalized and a mapping of service nodes onto server nodes is generated.

Claims

1. A method for generating automated mappings of service demands to server capacities in a distributed computer system, including a plurality of server and service nodes, the method comprising the steps of:

establishing at least one server model layer, said at least one server model layer including a set of server attribute values associated with the plurality of server nodes;

establishing at least one service model layer, said at least one service model layer including a set of service attribute values associated with the plurality of service nodes;

normalizing each set of server attribute values and service attribute values; and

generating a mapping of service nodes onto server nodes, using the normalized capacity attribute values and normalized demand attribute values.

2. The method of claim 1, wherein the method further comprises the steps of:

establishing a plurality of server model descriptions, such that the plurality of server model descriptions includes said plurality of server nodes; and

establishing a plurality of service model descriptions, such that the plurality of service model descriptions includes said plurality of service nodes.

3. The method of claim 1, wherein the method further comprises the steps of:

establishing a server domain which includes a plurality of server model layers, such that the server domain illustrates model layers that may include the services provided by software applications and software modules; and

establishing a service domain which includes a plurality of service model layers, such that the service model illustrates model layers that may include hardware that executes the software applications and modules in the service domain.

4. The method of claim 1, wherein the server nodes and service nodes include data structures, and the server nodes and service nodes illustrate the constituent elements of an individual model layer.

5. The method of claim 1, wherein the server attribute values and the service attribute values include capacity and demand attribute values, affinity and incentive constraints.

6. The method of claim 5, wherein the affinity constraints includes a continuum of values over a predetermined interval, such that there is an upper limit of the interval which represents absolute affinity and a lower limit of the interval which represents absolute repulsion.

7. The method of claim 5, wherein the incentive constraints include incentive functions, such that incentive functions are associated with non-technical considerations.

8. The method of claim 1, further comprising the step of:

performing a generic optimization process in order to produce an optimized mapping.

9. A method for allocating computing resources in a distributed computer system having a plurality of computing machines that host a plurality of application processes, the method comprising the steps of:

establishing at least one server model description, such that said at least one server model description includes data structures, and further includes a set of server attribute values for the plurality of server nodes;

establishing at least one service model description, such that said at least one service model description includes data structures, and further includes a set of service attribute values for the plurality of service nodes;

normalizing each of the associated sets of service and server attribute values by generating at least one value for each service and server attribute value, which represents an association between the plurality of service and server nodes; and

generating an optimized mapping of service nodes to server nodes.

10. The method according to claim 9, further comprising a service domain which includes a plurality of service model descriptions, describes the services provided by a software application and a software component that provides computing services;

and the server domain which includes a plurality of server model descriptions, describes the hardware that executes the software components in the service domain.

11. The method according to claim 9, wherein the server and service attribute values comprise a set of capacity and demand attributes values, affinity values and incentive functions, representing non-technical factors, and affinity values, which represent the degree of association between nodes.

12. The method according to claim 11, wherein the incentive value includes a constant value indicating the price of consuming one base unit of capacity and the incentive functions include variables which depend on parameters.

13. The method according to claim 11, wherein the affinity values include a continuum of values over a predetermined interval, such that there is an upper limit of the interval which represents absolute affinity and a lower limit of the interval which represents absolute repulsion.

14. The method according to claim 9, wherein the step of generating an optimized mapping further comprises the step of:

performing a generic optimization process to generate the optimized mapping, such that the generic optimization process takes into account constraints in the form of affinity-repulsion values and incentive values and functions.

15. The method according to claim 14, wherein the step of performing further comprises the steps of:

generating a estimated mapping based on a random mapping;

evaluating the estimated mapping according to the normalized constraints and a cost function;

replacing the estimated mapping with a poorer quality solution in a solution pool; and

repeating the steps until a predetermined termination criteria is met.

16. An apparatus for optimizing allocation of computer resources comprising:

server modeling means for generating a plurality of server nodes, wherein each of the plurality of server nodes has an associated set of server attribute values;

service modeling means for generating a plurality of service nodes, wherein each of the plurality of service nodes has an associated set of service attribute values;

normalizing means for normalizing the associated sets of server and service attribute values by generating at least one normalized value for each server and service attribute value, such that the normalized values describe predetermined relationships between the plurality of server and service nodes;

optimizing means for optimizing the normalized server and service attribute values in accordance with an objective function and an optimization function; and

mapping means for generating an optimized mapping of the demand model onto the capacity model.

17. The apparatus according to claim 16, wherein the server and service attribute values comprise capacity and demand constraints, incentive functions.

18. The apparatus according to claim 17, wherein the affinity values are a continuum of values over a predetermined interval, such that there is an upper limit of the interval which represents absolute affinity and a lower limit of the interval which represents absolute repulsion, and the incentive functions represent non-technical considerations.

19. The apparatus according to claim 17, wherein the affinity values are represented in an affinity/repulsion matrix and the incentive functions are represented as an incentive matrix.

20. The method according to claim 17, wherein the incentive functions indicate the price of consuming one base unit of capacity.

21. The apparatus according to claim 16, wherein the optimizing means includes means for performing a generic optimization process to generate an optimized mapping.

22. The apparatus according to claim 21, wherein the optimizing means further comprises:

selecting means for selecting one of a plurality of objective functions, wherein each of the plurality of the objective functions evaluates a solution as function of the demand and capacity attributes and formalized constraints.

23. The apparatus according to claim 21, wherein the optimizing means further comprises the steps of:

means for generating an estimated mapping based on a random mapping;

means for evaluating the estimated mapping according to the formalized constraints and a cost function;

means for replacing the estimated mapping with a poorer quality solution in a solution pool; and

means for repeating the steps until a predetermined termination criteria is met.

24. A computer readable storage medium on which is embedded one or more computer programs, said one or more computer programs implementing a method for generating automated mappings of service demands to server capacities in a distributed computer system, including a plurality of server and service nodes, said one or more computer programs comprising a set of instructions for:

establishing at least one server model layer, said at least one server model layer including a set of server attribute values associated with the plurality of server nodes;

establishing at least one service model layer, said at least one service model layer including a set of service attribute values associated with the plurality of service nodes;

normalizing each set of server attribute values and service attribute values; and

generating a mapping of service nodes onto server nodes, using the normalized capacity attribute values and normalized demand attribute values.

25. The computer readable storage medium according to claim 24, said one or more computer programs comprising a set of instructions for:

establishing a plurality of server model descriptions, such that the plurality of server model descriptions includes said plurality of server nodes; and

establishing a plurality of service model descriptions, such that the plurality of service model descriptions includes said plurality of service nodes.

26. The computer readable storage medium according to claim 24, said one or more computer programs comprising a set of instructions for:

establishing a server domain which includes a plurality of server model layers, such that the server domain illustrates model layers that may include the services provided by software applications and software modules; and

establishing a service domain which includes a plurality of service model layers, such that the service model illustrates model layers that may include hardware that executes the software applications and modules in the service domain.

27. The computer readable storage medium according to claim 24, wherein the server nodes and service nodes include data structures, and the server nodes and service nodes illustrate the constituent elements of an individual model layer.

28. The computer readable storage medium according to claim 24, wherein the server attribute values and the service attribute values include capacity and demand attribute values, affinity and incentive constraints.

29. The computer readable storage medium according to claim 28, wherein the affinity constraints includes a continuum of values over a predetermined interval, such that there is an upper limit of the interval which represents absolute affinity and a lower limit of the interval which represents absolute repulsion.

30. The computer readable storage medium according to claim 28, wherein the incentive constraints include incentive functions, such that incentive functions are associated with non-technical considerations.

31. The computer readable storage medium according to claim 24, said one or more computer programs comprising a set of instructions for:

performing a generic optimization process in order to produce an optimized mapping.