New! View global litigation for patent families

US20050197936A1 - Monte Carlo grid scheduling algorithm selection optimization - Google Patents

Monte Carlo grid scheduling algorithm selection optimization Download PDF

Info

Publication number
US20050197936A1
US20050197936A1 US10756112 US75611204A US2005197936A1 US 20050197936 A1 US20050197936 A1 US 20050197936A1 US 10756112 US10756112 US 10756112 US 75611204 A US75611204 A US 75611204A US 2005197936 A1 US2005197936 A1 US 2005197936A1
Authority
US
Grant status
Application
Patent type
Prior art keywords
scheduler
timetable
property
investment
return
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Abandoned
Application number
US10756112
Inventor
Viktors Berstis
Xiaolin Li
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
International Business Machines Corp
Original Assignee
International Business Machines Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management, e.g. organising, planning, scheduling or allocating time, human or machine resources; Enterprise planning; Organisational models
    • G06Q10/067Business modelling
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management, e.g. organising, planning, scheduling or allocating time, human or machine resources; Enterprise planning; Organisational models
    • G06Q10/063Operations research or analysis
    • G06Q10/0631Resource planning, allocation or scheduling for a business operation
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06QDATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00Finance; Insurance; Tax strategies; Processing of corporate or income taxes

Abstract

A method for utilizing the Monte Carlo method to determine the most efficient arrangement of schedulers for a grid using a Scheduler Optimization Program (SOP). The SOP obtains the schedulers and scheduler timetable from memory and randomly selects a time period and scheduler to analyze. The SOP then uses the selected scheduler to modify the scheduler timetable. The SOP then runs the ROI calculator to obtain a ROI property for the modified timetable. If the ROI property for the modified timetable is greater than the ROI property for the original scheduler timetable, the SOP replaces the scheduler timetable with the modified timetable.

Description

    CROSS-REFERENCE TO RELATED APPLICATION
  • [0001]
    The present invention is related to the subject matter of U.S. patent application ______ (Attorney docket number AUS920030689US1), which is incorporated herein by reference.
  • FIELD OF THE INVENTION
  • [0002]
    The present invention is related generally to methods of improving grid computing and specifically to an automated method for improving the selection of schedulers within a grid computing network.
  • BACKGROUND OF THE INVENTION
  • [0003]
    Recently, companies have begun to explore the possibilities of using grid computing networks (grids) to increase the company's productivity. A grid comprises a plurality of computers that are networked together. Large or complex computations (jobs) can be broken up into a plurality of smaller, more manageable jobs by the grid. The smaller jobs are then sent out to the computers within the grid for parallel processing. As the individual computers complete their jobs, the grid reassembles the smaller jobs into the completed job. The end result is that the large, complex jobs are processed in significantly less time than is possible on a single computer.
  • [0004]
    One of the important components of a grid is the scheduler. The scheduler is an algorithm that decides how to distribute the individual job pieces for processing throughout the grid. Although the concept of a scheduler sounds simple, the decision-making process for distribution of the job pieces is extremely complex. A number of decisions must be made as to how the scheduler chooses one grid computer over another. The physical distance between computers, processing speed, available memory, cost of operating grid computers, queue for each computer, the topology of the network connectivity among the computers, special resources (i.e. hardware, software, or licenses) available on particular computers, and connectivity between hard drives are just a few of the factors taken into consideration in creating a scheduler. Logical factors, such as job operating characteristics, priorities of various kinds, operational constraints on the utilization of the grid system, and many others, must also be taken into consideration by the scheduler. Thus, there are a plurality of different schedulers that can be created to distribute the job pieces throughout the grid. Selecting the appropriate scheduler for a grid is made even more complex by the fact that the type of grid traffic changes depending on the time of day or the day of the week, month, or year. One scheduler may be more efficient in the mornings and another scheduler may be more efficient in the evenings. A third scheduler may be more efficient on the weekends or on the last day of every fiscal quarter. Thus, in order to operate a grid at maximum efficiency, a user will ideally change schedulers from time to time, depending on the operating conditions of the grid. It is difficult for a person to constantly analyze and change the schedulers, so an automated method is preferable. Currently, there is no automated method for dynamically or adaptively changing the selection of the schedulers to best use at a given time interval. Consequently, a need exists for an automated method for dynamically and adaptively changing the selection of a scheduler in a grid computing network.
  • [0005]
    One method for measuring the efficiency of the scheduler is to run a return on investment (ROI) calculator. An example of an ROI calculator is described in U.S. patent application ______ (Attorney docket number AUS920030689US1) incorporated herein by reference. The ROI calculator can calculate, using simulation and other modeling methods, the return on investment of an IT infrastructure which employs a particular set of schedulers. The return on investment is a quantitative measure of how effectively the company's information technology (IT) infrastructure is implemented. The ROI calculator can also determine other properties associated with the grid such as operating efficiency, total operating cost, mean time to process individual jobs, and so forth. A user can run a ROI calculator for individual schedulers and operating conditions to determine which scheduler is best suited for which operating conditions. However, as schedulers are continuously modified and updated, an orderly method for applying the ROI calculator to the operating conditions and schedulers is needed. Therefore, what is needed is a method for determining how to select the operating conditions and schedulers using an ROI calculator as a measurement tool.
  • [0006]
    The Monte Carlo method for selecting criteria is well known in the art. The Monte Carlo method involves the random selection and application of criteria to a model. Proponents of the Monte Carlo method assert that the Monte Carlo method can be more efficient at finding near-optimum solutions than orderly search methods for particularly difficult problems. Schedulers and grid conditions, both being complex, are ideal for the Monte Carlo method. Therefore, what is needed is a method for applying the Monte Carlo method to schedulers and grid conditions for evaluation by an ROI calculator in order to determine the most efficient daily arrangement of schedulers to a grid.
  • SUMMARY OF THE INVENTION
  • [0007]
    The present invention, which meets the needs identified above, is a method for utilizing the Monte Carlo method to determine the most efficient arrangement of schedulers for a grid. The software embodiment of the present invention is a Scheduler Optimization Program (SOP). The SOP obtains the schedulers and scheduler timetable from memory and randomly selects a scheduler with which to modify the scheduler timetable at a randomly selected time period. The SOP assigns the randomly selected scheduler to the randomly selected time period in the scheduler timetable. The SOP compares the modified timetable to a previous results file to determine if the modified timetable was analyzed in a previous iteration. If the modified timetable was analyzed in a previous iteration, then SOP proceeds with another random selection as described above.
  • [0008]
    The SOP then runs the ROI calculator to obtain a ROI property for the modified timetable. The SOP then determines whether the ROI property for the modified timetable is greater than the ROI property for the original scheduler timetable. If the ROI property for the modified timetable is greater than the ROI property for the original scheduler timetable, then the SOP replaces the scheduler timetable with the modified timetable. The SOP repeats the iterative process described herein until a plateau is reached. The SOP may also be configured so that the SOP takes a configurable number of steps away from a more desirable ROI property (i.e. a local maximum) in an attempt to eventually reach a much more desirable ROI property (i.e. a regional or global maximum).
  • BRIEF DESCRIPTION OF THE DRAWINGS
  • [0009]
    The novel features believed characteristic of the invention are set forth in the appended claims. The invention itself, however, as well as a preferred mode of use, further objectives and advantages thereof, will best be understood by reference to the following detailed description of an illustrative embodiment when read in conjunction with the accompanying drawings, wherein:
  • [0010]
    FIG. 1 is an illustration of a computer network used to implement the present invention;
  • [0011]
    FIG. 2 is an illustration of a computer, including a memory and a processor, associated with the present invention;
  • [0012]
    FIG. 3 is an illustration of the logic of the Scheduler Optimization Program (SOP) of the present invention;
  • [0013]
    FIG. 4 is an illustration of the scheduler timetable of the present invention; and
  • [0014]
    FIG. 5 is an illustration of the previous results file of the present invention.
  • DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
  • [0015]
    As used herein, the term “computer” shall mean a machine having a processor, a memory, and an operating system, capable of interaction with a user or other computer, and shall include without limitation desktop computers, notebook computers, personal digital assistants (PDAs), servers, handheld computers, and similar devices.
  • [0016]
    As used herein, the term “efficiency” shall mean a quantitative measure of the amount of grid resources consumed to produce a desired effect.
  • [0017]
    As used herein, the term “modified timetable” shall mean a scheduler timetable in which the scheduler in a randomly selected time period has been replaced with a randomly selected scheduler.
  • [0018]
    As used herein, the term “previous results file” shall mean a computer file containing a list of modified schedulers that were analyzed in previous iterations of the present invention.
  • [0019]
    As used herein, the term “plateau” shall mean a state in which an iterative process no longer improves the ROI property.
  • [0020]
    As used herein, the term “ROI” is an acronym for return on investment.
  • [0021]
    As used herein, the term “ROI calculator” shall mean an algorithm for calculating a ROI property of a grid for a given time period and scheduler.
  • [0022]
    As used herein, the term “ROI property” shall mean a quantitative measure of a property for a grid computing network.
  • [0023]
    As used herein, the term “scheduler” shall mean a computer algorithm for determining the distribution of pieces of a job in a grid.
  • [0024]
    As used herein, the term “scheduler timetable” shall mean a table specifying the scheduler that a grid should use for a given time period.
  • [0025]
    As used herein, the term “time period” shall mean a specific block of time in a scheduler timetable or a modified timetable.
  • [0026]
    FIG. 1 is an illustration of computer network 90 associated with the present invention. Computer network 90 comprises local computer 95 electrically coupled to network 96. Local computer 95 is electrically coupled to remote computer 94 and remote computer 93 via network 96. Local computer 95 is also electrically coupled to server 91 and database 92 via network 96. Network 96 may be a simplified network connection such as a local area network (LAN) or may be a larger network such as a wide area network (WAN) or the Internet. Furthermore, computer network 90 depicted in FIG. 1 is intended as a representation of a possible operating network containing the present invention and is not meant as an architectural limitation.
  • [0027]
    The internal configuration of a computer, including connection and orientation of the processor, memory, and input/output devices, is well known in the art. The present invention is a methodology that can be embodied in a computer program. Referring to FIG. 2, the methodology of the present invention is implemented on software by Scheduler Optimization Program (SOP) 200. SOP 200 described herein can be stored within the memory of any computer depicted in FIG. 1. Alternatively, SOP 200 can be stored in an external storage device such as a removable disk, a CD-ROM, or a USB storage device. Memory 100 is illustrative of the memory within one of the computers of FIG. 1. Memory 100 also contains schedulers 120, scheduler timetable 140, previous results file 160, and return on investment (ROI) calculator 180. The present invention may interface with schedulers 120, scheduler timetable 140, previous results file 160, and ROI calculator 180 through memory 100. As part of the present invention, the memory 100 can be configured with SOP 200. Processor 106 can execute the instructions contained in SOP 200. Processor 106 is also able to display data on display 102 and accept user input on user input device 104. Processor 106, user input device 104, display 102, and memory 100 are part of a computer such as local computer 95 in FIG. 1. Processor 106 can communicate with other computers via network 96.
  • [0028]
    In alternative embodiments, SOP 200 can be stored in the memory of other computers. Storing SOP 200 in the memory of other computers allows the processor workload to be distributed across a plurality of processors instead of a single processor. Further configurations of SOP 200 across various memories are known by persons of ordinary skill in the art. The present invention may be a method, a stand alone computer program, or a plug-in to an existing computer program. Persons of ordinary skill in the art are aware of how to configure computer programs, such as those described herein, to plug into an existing computer program.
  • [0029]
    Schedulers 120 are computer algorithms that decide where the pieces of a job are distributed throughout the grid. For the purposes herein, schedulers 120 are represented as scheduler A, scheduler B, scheduler C and so forth. Persons of ordinary skill in the art will appreciate that each letter represents a scheduler with a particular configuration. Scheduler timetable 140 is a computer file that tells the grid computing system which scheduler to run at which times. Scheduler timetable 140 may be for a day, a week, a month, a year, or any other time period as determined by a person of ordinary skill in the art. Previous results file 160 is a listing of the modified timetables that were previously selected by SOP 200. ROI calculator 180 contains a mathematical model of the operating conditions within the grid. ROI calculator 180 calculates a ROI property for the modified scheduler. The ROI properties include the time for return on initial investment, the annual operating cost savings, the efficiency with which the grid is being utilized, and so forth. Persons skilled in the art are aware of other types of calculators that can determine ROI properties.
  • [0030]
    FIG. 3 illustrates the logic of Scheduler Optimization Program (SOP) 200 of the present invention. SOP 200 is a program that uses the Monte Carlo method to optimize the selection of schedulers 120 for the grid computing system. SOP 200 starts (202) anytime a user desires to optimize scheduler timetable 140 of the present invention. SOP 200 then obtains schedulers 120 and scheduler timetable 140 from memory (204). Scheduler timetable 140 contains the ROI property for the current arrangement of schedulers 120 for a specific time period, such as a single day. SOP 200 then randomly selects a time period to modify in scheduler timetable 140 (206). The time period is a specific block of time within scheduler timetable 140.50P 200 then randomly selects a scheduler to replace the original scheduler in a randomly selected time period within scheduler timetable 140 (208). SOP 200 selects the scheduler from the plurality of schedulers 120.
  • [0031]
    SOP 200 then creates a modified timetable by inserting the randomly selected scheduler into the randomly selected time period in the scheduler timetable (210). SOP 200 then determines whether the modified timetable created in step 210 is in previous results file 160 (212). By checking the modified timetable against previous results file 160, the present invention does not run ROI calculator 180 when the present invention has previously run ROI calculator 180 for the same modified timetable in a previous iteration of SOP 200. If the modified timetable is in previous results file 160, then SOP 200 returns to step 206. If the modified timetable is not in previous results file 160, then SOP 200 saves the modified timetable in previous results file 160 (214). SOP 200 then runs ROI calculator 180 to obtain an ROI property for the modified timetable (216). For the illustrative purposes herein, the ROI property is the grid efficiency. Persons of ordinary skill in the art are aware that any of the ROI properties calculated by ROI calculator 180 can be used to optimize the selection of schedulers 120 herein.
  • [0032]
    At step 218, SOP 200 determines whether the ROI property for the modified timetable is greater than the ROI property for the original scheduler timetable 140 (218). The term “greater than” in step 218 is meant to mean more desirable. If the ROI property is the grid efficiency, then at step 218 SOP 200 determines whether the grid efficiency for the modified timetable is greater than the grid efficiency for the original scheduler timetable 140. If the grid efficiency for the modified timetable is not greater than the grid efficiency for the original scheduler timetable 140, then SOP 200 proceeds to step 232. If the grid efficiency for the modified timetable is greater than the grid efficiency for the original scheduler timetable 140, then SOP 200 replaces the original scheduler timetable 140 with the modified timetable (220). SOP 200 also replaces the original ROI property with the ROI property for the modified timetable calculated in step 216. Persons of ordinary skill in the art will appreciate that SOP 200 could also rank the different scheduler timetables 140 and/or modified timetables according to their ROI property and run the modification process described herein on the top N number of scheduler timetables 140 and/or modified timetables in the ranked list, wherein N is a user-configurable number. The present may also be embodied such that a random scheduler is selected to replace a scheduler in a random time period in a randomly selected scheduler timetable 140 or modified timetable in the ranked list.
  • [0033]
    SOP 200 then determines whether a plateau has been reached for scheduler timetable 140 (222). A plateau may be reached whenever SOP 200 has not modified scheduler timetable 140 in a certain number (i.e. one thousand) of iterations. Alternatively, a user of the present invention may define when a plateau is reached, such as when scheduler timetable 140 has not produced an improved ROI property after a configurable number of attempts or has an acceptable ROI property. Further in the alternative, a user of the present invention may choose to stop the iterative process of SOP 200 by manually indicating that a plateau has been reached. If the present invention is performing the modification process on a ranked list of scheduler timetables 140 and/or modified timetables, then a plateau may be defined as a user-configurable number of attempts in which the top N timetables have not changed, N being a user-configurable number. Persons of ordinary skill in the art are aware of other methods of determining if a plateau has been reached. If a plateau has not been reached, SOP 200 returns to step 206. If a plateau has been reached, SOP 200 proceeds to step 224.
  • [0034]
    Steps 202 through 222 of SOP 200 of the present invention continuously seek to improve the grid computing network by improving the ROI property. However, the present invention may be configured such that SOP 200 takes a user-configurable number of steps towards a less desirable result in an effort to ultimately achieve a more desirable result. As an analogy, if a person only goes uphill, then the person will make it to the top of the hill he is on, but he will never reach the top of the highest hill in the area (unless he happens to have climbed the highest hill in the area the first time). In order to reach the top of the highest hill in the area, the person must descend and traverse a valley before climbing a new hill. In order to retain the ability to achieve a possibly superior result, a person of ordinary skill in the art will appreciate that SOP 200 can be configured so that SOP 200 takes a user-configurable number of steps towards an undesirable result in order to possibly achieve a much more desirable result. Steps 224 through 240 of SOP 200 illustrate the process of taking steps towards the undesirable result.
  • [0035]
    At step 224, SOP 200 determines if the user has indicated a desire to take steps away from the plateau (224). If the user has indicated a desire to take steps away from the plateau, the user will also specify how many steps to take away from the plateau. If the user has not indicated a desire to take steps away from the plateau, SOP 200 ends (226). If the user has indicated a desire to take steps away from the plateau, then SOP 200 randomly selects a time period to modify in scheduler timetable 140 (228). SOP 200 then randomly selects a scheduler to replace the original scheduler in a randomly selected time period within scheduler timetable 140 (230). SOP 200 then creates a modified timetable by inserting the randomly selected scheduler into the randomly selected time period in the scheduler timetable (232). SOP 200 then runs ROI calculator 180 to obtain an ROI property for the modified timetable (234), and proceeds to step 236.
  • [0036]
    At step 236, SOP 200 determines whether the ROI property for the modified timetable is greater than the ROI property for the original scheduler timetable 140 (236). If the ROI property for the modified timetable is greater than the ROI property for the original scheduler timetable 140, then SOP 200 proceeds to step 240. If the grid efficiency for the modified timetable is not greater than the grid efficiency for the original scheduler timetable 140, then SOP 200 replaces the original scheduler timetable 140 with the modified timetable (238). SOP 200 also replaces the original ROI property with the ROI property for the modified timetable calculated in step 234. SOP 200 then determines if SOP 200 has reached the user-configured number of backwards steps (240). If SOP 200 has not reached the user-configured number of backwards steps, SOP 200 returns to step 228. If SOP 200 has reached the user-configured number of backwards steps, SOP 200 returns to step 206. If desired, the results from the actual grid operation using a particular scheduler can be saved and compared against the ROI calculator's 180 estimate. Adjustments can then be made to ROI calculator's 180 template in order to improve the accuracy of ROI calculator 180. Persons skilled in the art will recognize that steps 224 through 240 may be included in SOP 200 prior to a plateau being reached and, if so, these steps may be implemented on a random basis. In addition, the user configurable number of steps may be replaced by a randomly determined number of steps.
  • [0037]
    SOP 200 of the present invention modifies scheduler timetable 140 with a new scheduler 120 and calculates the ROI property for the entire modified scheduler. It is possible that a similar result may be achieved by dividing the scheduler timetable 140 into a plurality of short time periods, each with one scheduler 120, and running the ROI calculator on the individual scheduler timetable pieces. However, the described embodiment is preferable to an embodiment in which the scheduler timetable 140 is divided into a plurality of short time periods because the selection of one scheduler 120 affects the performance of subsequent schedulers 120. For example, scheduler B may be very efficient at processing memory intensive jobs. If scheduler A is operating before scheduler B and prioritizes memory intensive jobs, then when scheduler B comes online, there are relatively few memory intensive jobs and scheduler B's efficiency is low. Conversely, if scheduler C is operating before scheduler B and prioritizes processor intensive jobs, then when scheduler B comes online, there may be many memory intensive jobs and scheduler B's efficiency is high. Thus, the performance of any one scheduler 120 is dependent on the type of jobs that remain after the previous scheduler 120 is taken offline. Therefore, the evaluation of scheduler timetable 140 as a whole is preferable.
  • [0038]
    FIG. 4 illustrates scheduler timetable 140 of the present invention. Scheduler timetable 140 comprises three rows: time 142, scheduler 144, and ROI property 146. Time 142 is the specific time period of implementation of scheduler 120. Scheduler 144 indicates which specific scheduler 120 will be implemented at the appropriate time 142. ROI property 146 is the ROI property that is used in step 218 of SOP 200. In FIG. 4, ROI property 146 is the grid network efficiency.
  • [0039]
    FIG. 5 illustrates previous results file 160. Previous results file 160 comprises modified timetable ID 161, time 162, and scheduler 164. Modified timetable ID 161 is a counter used to distinguish individual modified timetables from each other. Time 162 and scheduler 164 are the components of a modified timetable, similar to scheduler timetable 140 depicted in FIG. 4. Time 162 is the time period for the modified timetable. Time 162 is like time 142 in FIG. 4. Scheduler 164 is scheduler 120 for the modified timetable. Scheduler 164 is like scheduler 144 in FIG. 4.
  • [0040]
    With respect to the above description, it is to be realized that the optimum dimensional relationships for the parts of the invention, to include variations in size, materials, shape, form, function, manner of operation, assembly, and use are deemed readily apparent and obvious to one of ordinary skill in the art. The present invention encompasses all equivalent relationships to those illustrated in the drawings and described in the specification. The novel spirit of the present invention is still embodied by reordering or deleting some of the steps contained in this disclosure. The spirit of the invention is not meant to be limited in any way except by proper construction of the following claims.

Claims (31)

  1. 1. A method for optimizing an allocation of a plurality of schedulers within a grid computing network wherein the allocation is determined by a return on investment calculator.
  2. 2. A method of claim 1 comprising:
    selecting a first scheduler;
    obtaining a first scheduler timetable and a first scheduler timetable return on investment property;
    creating a second scheduler timetable by modifying the first scheduler timetable with the first scheduler;
    running the return on investment calculator to obtain a second scheduler return on investment property;
    determining whether the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property; and
    responsive to the determination that the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property, replacing the first scheduler timetable with the second scheduler timetable.
  3. 3. The method of claim 2 further comprising:
    responsive to the determination that the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property, replacing the first scheduler timetable return on investment property with the second scheduler timetable return on investment property.
  4. 4. The method of claim 2 further comprising:
    determining whether the second scheduler timetable is in a previous results file; and
    responsive to the determination that the second scheduler timetable is not in the previous results file, saving the second scheduler timetable in the previous results file and performing the steps in claim 2.
  5. 5. The method of claim 2 further comprising:
    determining whether the second scheduler timetable is in a previous results file; and
    responsive to the determination that the second scheduler timetable is in the previous results file, selecting a new first scheduler.
  6. 6. The method of claim 2 further comprising:
    determining if a plateau has been reached; and
    responsive to the determination that the plateau has not been reached, repeating the steps of claim 2.
  7. 7. The method of claim 2 wherein the selection of a first time period and the first scheduler is random.
  8. 8. The method of claim 2 further comprising: taking a user configurable number of steps away from a greater return on investment property, and then repeating the steps in claim 2.
  9. 9. The method of claim 8 wherein the step of taking a user configurable number of steps away from a greater return on investment property comprises:
    accepting a user-configurable number of backward steps;
    taking a backward step comprising:
    selecting a first scheduler;
    obtaining a first scheduler timetable and a first scheduler timetable return on investment property;
    creating a second scheduler timetable by modifying the first scheduler timetable with the first scheduler;
    running the return on investment calculator to obtain a second scheduler return on investment property;
    determining whether the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property; and
    responsive to the determination that the second scheduler timetable return on investment property is not greater than the first scheduler timetable return on investment property, replacing the first scheduler timetable with the second scheduler timetable.
  10. 10. The method of claim 9 wherein taking the backward step further comprising:
    responsive to the determination that the second scheduler timetable return on investment property is not greater than the first scheduler timetable return on investment property, replacing the first scheduler timetable return on investment property with the second scheduler timetable return on investment property.
  11. 11. The method of claim 9 further comprising:
    determining whether the number of backward steps taken equals the user-configurable number of backward steps;
    responsive to the determination that the number of backward steps taken equals the user-configurable number of backward steps, repeating the steps in claim 2.
  12. 12. The method of claim 9 further comprising:
    determining whether the number of backward steps taken equals the user-configurable number of backward steps;
    responsive to the determination that the number of backward steps taken does not equal the user-configurable number of backward steps, repeating the steps in claim 9.
  13. 13. A program product for optimizing an allocation of a plurality of schedulers within a grid computing network wherein the allocation is determined by a return on investment calculator.
  14. 14. A program product of claim 13 comprising:
    a computer-usable medium;
    wherein the computer usable medium comprises instructions comprising:
    instructions for selecting a first scheduler;
    instructions for obtaining a first scheduler timetable and a first scheduler timetable return on investment property;
    instructions for creating a second scheduler timetable by modifying the first scheduler timetable with the first scheduler;
    instructions for running the return on investment calculator to obtain a second scheduler return on investment property;
    instructions for determining whether the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property; and
    responsive to the determination that the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property, instructions for replacing the first scheduler timetable with the second scheduler timetable.
  15. 15. The program product of claim 14 further comprising:
    responsive to the determination that the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property, instructions for replacing the first scheduler timetable return on investment property with the second scheduler timetable return on investment property.
  16. 16. The program product of claim 14 further comprising:
    instructions for determining whether the second scheduler timetable is in a previous results file; and
    responsive to the determination that the second scheduler timetable is not in the previous results file, instructions for saving the second scheduler timetable in the previous results file and performing the steps in claim 14.
  17. 17. The program product of claim 14 further comprising:
    instructions for determining whether the second scheduler timetable is in a previous results file; and
    responsive to the determination that the second scheduler timetable is in the previous results file, instructions for selecting a new first scheduler.
  18. 18. The program product of claim 14 further comprising:
    instructions for determining if a plateau has been reached; and
    responsive to the determination that the plateau has not been reached, instructions for repeating the steps of claim 14.
  19. 19. The program product of claim 14 wherein the selection of a first time period and the first scheduler is random.
  20. 20. The program product of claim 14 further comprising: taking a user configurable number of steps away from a greater return on investment property, and then repeating the steps in claim 14.
  21. 21. The program product of claim 20 wherein the step of taking a user configurable number of steps away from a greater return on investment property comprises:
    instructions for accepting a user-configurable number of backward steps;
    instructions for taking a backward step comprising:
    instructions for selecting a first scheduler;
    instructions for obtaining a first scheduler timetable and a first scheduler timetable return on investment property;
    instructions for creating a second scheduler timetable by modifying the first scheduler timetable with the first scheduler;
    instructions for running the return on investment calculator to obtain a second scheduler return on investment property;
    instructions for determining whether the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property; and
    responsive to the determination that the second scheduler timetable return on investment property is not greater than the first scheduler timetable return on investment property, instructions for replacing the first scheduler timetable with the second scheduler timetable.
  22. 22. The program product of claim 21 wherein taking the backward step further comprising:
    responsive to the determination that the second scheduler timetable return on investment property is not greater than the first scheduler timetable return on investment property, instructions for replacing the first scheduler timetable return on investment property with the second scheduler timetable return on investment property.
  23. 23. The program product of claim 21 further comprising:
    instructions for determining whether the number of backward steps taken equals the user-configurable number of backward steps;
    responsive to the determination that the number of backward steps taken equals the user-configurable number of backward steps, instructions for repeating the steps in claim 14.
  24. 24. The program product of claim 21 further comprising:
    instructions for determining whether the number of backward steps taken equals the user-configurable number of backward steps;
    responsive to the determination that the number of backward steps taken does not equal the user-configurable number of backward steps, instructions for repeating the steps in claim 21.
  25. 25. An apparatus for optimizing an allocation of a plurality of schedulers within a grid computing network wherein the allocation is determined by a return on investment calculator comprising:
    means for selecting a first scheduler;
    means for obtaining a first scheduler timetable and a first scheduler timetable return on investment property;
    means for creating a second scheduler timetable by modifying the first scheduler timetable with the first scheduler;
    means for selecting a first time period;
    means for determining whether the first time period coincides with a second time period in the first scheduler timetable;
    responsive to the determination that the first time period does not coincide with the second time period, means for dividing the first time period and the second time period so that the first time period and the second time period coincide;
    means for determining whether the second scheduler timetable is in a previous results file; and
    responsive to the determination that the second scheduler timetable is in the previous results file, means for selecting a new first scheduler;
    responsive to the determination that the second scheduler timetable is not in the previous results file, means for saving the second scheduler timetable in the previous results file;
    means for running the return on investment calculator to obtain a second scheduler return on investment property;
    means for determining whether the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property;
    responsive to the determination that the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property, means for replacing the first scheduler timetable with the second scheduler timetable;
    responsive to the determination that the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property, means for replacing the first scheduler timetable return on investment property with the second scheduler timetable return on investment property; and
    wherein the selection of a first time period and the first scheduler is random.
  26. 26. The apparatus of claim 17 further comprising:
    means for determining if a plateau has been reached; and
    responsive to the determination that the plateau has not been reached, means for repeating the steps of claim 17.
  27. 27. The apparatus of claim 17 further comprising: means for taking a user configurable number of steps away from a greater return on investment property, and then repeating the steps in claim 17.
  28. 28. The apparatus of claim 27 wherein the step of taking a user configurable number of steps away from a greater return on investment property comprises:
    means for accepting a user-configurable number of backward steps;
    means for taking a backward step comprising:
    means for selecting a first scheduler;
    means for obtaining a first scheduler timetable and a first scheduler timetable return on investment property;
    means for creating a second scheduler timetable by modifying the first scheduler timetable with the first scheduler;
    means for running the return on investment calculator to obtain a second scheduler return on investment property;
    means for determining whether the second scheduler timetable return on investment property is greater than the first scheduler timetable return on investment property; and
    responsive to the determination that the second scheduler timetable return on investment property is not greater than the first scheduler timetable return on investment property, means for replacing the first scheduler timetable with the second scheduler timetable.
  29. 29. The apparatus of claim 28 wherein taking the backward step further comprising:
    responsive to the determination that the second scheduler timetable return on investment property is not greater than the first scheduler timetable return on investment property, means for replacing the first scheduler timetable return on investment property with the second scheduler timetable return on investment property.
  30. 30. The apparatus of claim 28 further comprising:
    means for determining whether the number of backward steps taken equals the user-configurable number of backward steps;
    responsive to the determination that the number of backward steps taken equals the user-configurable number of backward steps, means for repeating the steps in claim 25.
  31. 31. The apparatus of claim 28 further comprising:
    means for determining whether the number of backward steps taken equals the user-configurable number of backward steps;
    responsive to the determination that the number of backward steps taken does not equal the user-configurable number of backward steps, means for repeating the steps in claim 28.
US10756112 2004-01-13 2004-01-13 Monte Carlo grid scheduling algorithm selection optimization Abandoned US20050197936A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US10756112 US20050197936A1 (en) 2004-01-13 2004-01-13 Monte Carlo grid scheduling algorithm selection optimization

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US10756112 US20050197936A1 (en) 2004-01-13 2004-01-13 Monte Carlo grid scheduling algorithm selection optimization
US12174755 US8024209B2 (en) 2004-01-13 2008-07-17 Monte carlo grid scheduling algorithm selection optimization

Publications (1)

Publication Number Publication Date
US20050197936A1 true true US20050197936A1 (en) 2005-09-08

Family

ID=34911234

Family Applications (2)

Application Number Title Priority Date Filing Date
US10756112 Abandoned US20050197936A1 (en) 2004-01-13 2004-01-13 Monte Carlo grid scheduling algorithm selection optimization
US12174755 Expired - Fee Related US8024209B2 (en) 2004-01-13 2008-07-17 Monte carlo grid scheduling algorithm selection optimization

Family Applications After (1)

Application Number Title Priority Date Filing Date
US12174755 Expired - Fee Related US8024209B2 (en) 2004-01-13 2008-07-17 Monte carlo grid scheduling algorithm selection optimization

Country Status (1)

Country Link
US (2) US20050197936A1 (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080275804A1 (en) * 2004-01-13 2008-11-06 Viktors Berstis Monte Carlo Grid Scheduling Algorithm Selection Optimization
US20110035247A1 (en) * 2009-08-04 2011-02-10 United Parcel Service Of America Systems, methods, apparatuses, and computer program products for determining productivity associated with retrieving items in a warehouse
US20130132962A1 (en) * 2011-11-22 2013-05-23 Microsoft Corporation Scheduler combinators
WO2014142498A1 (en) * 2013-03-12 2014-09-18 삼성전자 주식회사 Method and system for scheduling computing
US20160110812A1 (en) * 2012-12-18 2016-04-21 Johnathan Mun Project economics analysis tool

Citations (19)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5222192A (en) * 1988-02-17 1993-06-22 The Rowland Institute For Science, Inc. Optimization techniques using genetic algorithms
US5241465A (en) * 1990-04-27 1993-08-31 Hitachi Ltd. Method for determining optimum schedule in computer-aided scheduling system
US5414845A (en) * 1992-06-26 1995-05-09 International Business Machines Corporation Network-based computer system with improved network scheduling system
US5524077A (en) * 1987-07-24 1996-06-04 Faaland; Bruce H. Scheduling method and system
US5590063A (en) * 1994-07-05 1996-12-31 Motorola, Inc. Optimization method using parallel processors
US5848403A (en) * 1996-10-04 1998-12-08 Bbn Corporation System and method for genetic algorithm scheduling systems
US6004015A (en) * 1994-11-24 1999-12-21 Matsushita Electric Industrial Co., Ltd. Optimization adjusting method and optimization adjusting apparatus
US6032172A (en) * 1997-05-29 2000-02-29 Sun Microsystems, Inc. Distributed computer process scheduling mechanism
US6035278A (en) * 1997-07-08 2000-03-07 Netscape Communications Corporation Method and system for schedule and task management
US6289296B1 (en) * 1997-04-01 2001-09-11 The Institute Of Physical And Chemical Research (Riken) Statistical simulation method and corresponding simulation system responsive to a storing medium in which statistical simulation program is recorded
US6381586B1 (en) * 1998-12-10 2002-04-30 International Business Machines Corporation Pricing of options using importance sampling and stratification/ Quasi-Monte Carlo
US20020143590A1 (en) * 2001-03-27 2002-10-03 Dhong Sang Hoo Method and apparatus for evaluating results of multiple software tools
US20020184069A1 (en) * 2001-05-17 2002-12-05 Kosiba Eric D. System and method for generating forecasts and analysis of contact center behavior for planning purposes
US6578005B1 (en) * 1996-11-22 2003-06-10 British Telecommunications Public Limited Company Method and apparatus for resource allocation when schedule changes are incorporated in real time
US20030177060A1 (en) * 2002-03-12 2003-09-18 Seagraves Theresa L. System and method for return on investment
US20040078310A1 (en) * 2002-10-17 2004-04-22 Louis Shaffer System and method for determining a return-on-investment in a semiconductor or data storage fabrication facility
US6823315B1 (en) * 1999-11-03 2004-11-23 Kronos Technology Systems Limited Partnership Dynamic workforce scheduler
US6882989B2 (en) * 2001-02-23 2005-04-19 Bbnt Solutions Llc Genetic algorithm techniques and applications
US7246075B1 (en) * 2000-06-23 2007-07-17 North Carolina A&T State University System for scheduling multiple time dependent events

Family Cites Families (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5050213A (en) 1986-10-14 1991-09-17 Electronic Publishing Resources, Inc. Database usage metering and protection system and method
DE69122779D1 (en) 1991-11-19 1996-11-21 Pmb Holding Bv fürzigarren wrapping
US5390281A (en) 1992-05-27 1995-02-14 Apple Computer, Inc. Method and apparatus for deducing user intent and providing computer implemented services
US6278978B1 (en) * 1998-04-07 2001-08-21 Blue Pumpkin Software, Inc. Agent scheduling system and method having improved post-processing step
US6338149B1 (en) 1998-07-31 2002-01-08 Westinghouse Electric Company Llc Change monitoring system for a computer system
US6249769B1 (en) 1998-11-02 2001-06-19 International Business Machines Corporation Method, system and program product for evaluating the business requirements of an enterprise for generating business solution deliverables
US6418462B1 (en) 1999-01-07 2002-07-09 Yongyong Xu Global sideband service distributed computing method
US6442164B1 (en) 1999-06-03 2002-08-27 Fujitsu Network Communications, Inc. Method and system for allocating bandwidth and buffer resources to constant bit rate (CBR) traffic
US7640547B2 (en) * 2002-02-08 2009-12-29 Jpmorgan Chase & Co. System and method for allocating computing resources of a distributed computing system
US7191435B2 (en) 2002-06-07 2007-03-13 Sun Microsystems, Inc. Method and system for optimizing software upgrades
US7395235B2 (en) * 2002-06-13 2008-07-01 Centre For Development Of Advanced Computing Strategy independent optimization of multi objective functions
JP4069701B2 (en) 2002-07-31 2008-04-02 株式会社日立製作所 Work support apparatus
JP4310090B2 (en) 2002-09-27 2009-08-05 株式会社日立ハイテクノロジーズ Defect data analyzing method and apparatus, and review system
US20050197936A1 (en) 2004-01-13 2005-09-08 International Business Machines Corporation Monte Carlo grid scheduling algorithm selection optimization
US7603253B2 (en) 2004-01-13 2009-10-13 International Business Machines Corporation Apparatus and method for automatically improving a set of initial return on investment calculator templates

Patent Citations (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5524077A (en) * 1987-07-24 1996-06-04 Faaland; Bruce H. Scheduling method and system
US5222192A (en) * 1988-02-17 1993-06-22 The Rowland Institute For Science, Inc. Optimization techniques using genetic algorithms
US5241465A (en) * 1990-04-27 1993-08-31 Hitachi Ltd. Method for determining optimum schedule in computer-aided scheduling system
US5414845A (en) * 1992-06-26 1995-05-09 International Business Machines Corporation Network-based computer system with improved network scheduling system
US5590063A (en) * 1994-07-05 1996-12-31 Motorola, Inc. Optimization method using parallel processors
US6148274A (en) * 1994-11-24 2000-11-14 Matsushita Electric Industrial Co., Ltd. Optimization adjusting method and optimization adjusting apparatus
US6004015A (en) * 1994-11-24 1999-12-21 Matsushita Electric Industrial Co., Ltd. Optimization adjusting method and optimization adjusting apparatus
US5848403A (en) * 1996-10-04 1998-12-08 Bbn Corporation System and method for genetic algorithm scheduling systems
US6578005B1 (en) * 1996-11-22 2003-06-10 British Telecommunications Public Limited Company Method and apparatus for resource allocation when schedule changes are incorporated in real time
US6289296B1 (en) * 1997-04-01 2001-09-11 The Institute Of Physical And Chemical Research (Riken) Statistical simulation method and corresponding simulation system responsive to a storing medium in which statistical simulation program is recorded
US6032172A (en) * 1997-05-29 2000-02-29 Sun Microsystems, Inc. Distributed computer process scheduling mechanism
US6035278A (en) * 1997-07-08 2000-03-07 Netscape Communications Corporation Method and system for schedule and task management
US6381586B1 (en) * 1998-12-10 2002-04-30 International Business Machines Corporation Pricing of options using importance sampling and stratification/ Quasi-Monte Carlo
US6823315B1 (en) * 1999-11-03 2004-11-23 Kronos Technology Systems Limited Partnership Dynamic workforce scheduler
US7246075B1 (en) * 2000-06-23 2007-07-17 North Carolina A&T State University System for scheduling multiple time dependent events
US6882989B2 (en) * 2001-02-23 2005-04-19 Bbnt Solutions Llc Genetic algorithm techniques and applications
US20020143590A1 (en) * 2001-03-27 2002-10-03 Dhong Sang Hoo Method and apparatus for evaluating results of multiple software tools
US20020184069A1 (en) * 2001-05-17 2002-12-05 Kosiba Eric D. System and method for generating forecasts and analysis of contact center behavior for planning purposes
US20030177060A1 (en) * 2002-03-12 2003-09-18 Seagraves Theresa L. System and method for return on investment
US20040078310A1 (en) * 2002-10-17 2004-04-22 Louis Shaffer System and method for determining a return-on-investment in a semiconductor or data storage fabrication facility

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080275804A1 (en) * 2004-01-13 2008-11-06 Viktors Berstis Monte Carlo Grid Scheduling Algorithm Selection Optimization
US8024209B2 (en) 2004-01-13 2011-09-20 International Business Machines Corporation Monte carlo grid scheduling algorithm selection optimization
US20110035247A1 (en) * 2009-08-04 2011-02-10 United Parcel Service Of America Systems, methods, apparatuses, and computer program products for determining productivity associated with retrieving items in a warehouse
US20130132962A1 (en) * 2011-11-22 2013-05-23 Microsoft Corporation Scheduler combinators
US20160110812A1 (en) * 2012-12-18 2016-04-21 Johnathan Mun Project economics analysis tool
US9881339B2 (en) * 2012-12-18 2018-01-30 Johnathan Mun Project economics analysis tool
WO2014142498A1 (en) * 2013-03-12 2014-09-18 삼성전자 주식회사 Method and system for scheduling computing

Also Published As

Publication number Publication date Type
US20080275804A1 (en) 2008-11-06 application
US8024209B2 (en) 2011-09-20 grant

Similar Documents

Publication Publication Date Title
Lublin et al. The workload on parallel supercomputers: modeling the characteristics of rigid jobs
Wolf et al. Flex: A slot allocation scheduling optimizer for mapreduce workloads
Albers et al. Speed scaling on parallel processors
Smit et al. Comparing parameter tuning methods for evolutionary algorithms
Bansal et al. Speed scaling for weighted flow time
Carman et al. Towards an economy-based optimisation of file access and replication on a data grid
Van den Bossche et al. Online cost-efficient scheduling of deadline-constrained workloads on hybrid clouds
US7668703B1 (en) Determining required capacity for a resource
US20110161964A1 (en) Utility-Optimized Scheduling of Time-Sensitive Tasks in a Resource-Constrained Environment
Malawski et al. Algorithms for cost-and deadline-constrained provisioning for scientific workflow ensembles in IaaS clouds
US20050262506A1 (en) Grid non-deterministic job scheduling
US20100005173A1 (en) Method, system and computer program product for server selection, application placement and consolidation
US20090031312A1 (en) Method and Apparatus for Scheduling Grid Jobs Using a Dynamic Grid Scheduling Policy
US20140215487A1 (en) Optimizing execution and resource usage in large scale computing
Ganapathi et al. Statistics-driven workload modeling for the cloud
US20130139152A1 (en) Cloud provisioning accelerator
US20070256077A1 (en) Fair share scheduling based on an individual user's resource usage and the tracking of that usage
Ghosh et al. Modeling and performance analysis of large scale iaas clouds
Li et al. Energy-efficient stochastic task scheduling on heterogeneous computing systems
US20090113434A1 (en) Apparatus, system and method for rapid resource scheduling in a compute farm
Xhafa et al. Batch mode scheduling in grid systems
US20070294408A1 (en) Optimized multi-component co-allocation scheduling with advanced reservations for data transfers and distributed jobs
Verma et al. Two sides of a coin: Optimizing the schedule of mapreduce jobs to minimize their makespan and improve cluster performance
Wirtz et al. Improving mapreduce energy efficiency for computation intensive workloads
US20130318538A1 (en) Estimating a performance characteristic of a job using a performance model

Legal Events

Date Code Title Description
AS Assignment

Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:BERSTIS, VIKTORS;LI, XIAOLIN;REEL/FRAME:014936/0066;SIGNING DATES FROM 20031204 TO 20031210