CA2558826C - Dependency graph parameter scoping - Google Patents

Dependency graph parameter scoping Download PDF

Info

Publication number
CA2558826C
CA2558826C CA2558826A CA2558826A CA2558826C CA 2558826 C CA2558826 C CA 2558826C CA 2558826 A CA2558826 A CA 2558826A CA 2558826 A CA2558826 A CA 2558826A CA 2558826 C CA2558826 C CA 2558826C
Authority
CA
Canada
Prior art keywords
tasks
parameter
value
task
identifying
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.)
Active
Application number
CA2558826A
Other languages
French (fr)
Other versions
CA2558826A1 (en
Inventor
Frank Inchingolo
Craig W. Stanfill
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.)
Ab Initio Technology LLC
Original Assignee
Ab Initio Technology LLC
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
Application filed by Ab Initio Technology LLC filed Critical Ab Initio Technology LLC
Priority to CA2880884A priority Critical patent/CA2880884C/en
Publication of CA2558826A1 publication Critical patent/CA2558826A1/en
Application granted granted Critical
Publication of CA2558826C publication Critical patent/CA2558826C/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F8/00Arrangements for software engineering
    • G06F8/40Transformation of program code
    • G06F8/41Compilation
    • G06F8/43Checking; Contextual analysis
    • G06F8/433Dependency analysis; Data or control flow analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/455Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45504Abstract machines for programme code execution, e.g. Java virtual machine [JVM], interpreters, emulators
    • G06F9/45508Runtime interpretation or emulation, e g. emulator loops, bytecode interpretation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling

Abstract

A number of tasks are defined according to a dependency graph. Multiple parameter contexts are maintained, each associated with a different scope of the tasks. A parameter used in a first of the tasks is bound to a value. This binding includes identifying a first of the contexts according to the dependency graph and retrieving the value for the parameter from the identified context.

Description

DEPENDENCY GRAPH PARAMETER SCOPING
Background This invention relates to scoping of parameters in a dependency graph.
Computer job control systems have been used to sequence execution of computer programs or other processes according to constraints related to the order in which the programs are to be executed. One approach to specifying the ordering constraints is using a dependency graph. The programs may accept arguments that are specified at the time jobs are submitted to the system. One example of such job control systems is a distributed batch systems in which different jobs can be specified to execute on different computers or types of computers (e.g., supercomputers), and dependencies between different jobs are explicitly identified before the jobs are executed. One approach to communication of information between jobs is through a common data store such as through files in a file system. Another example of such job control relates to job scheduling in a distributed personal computer environment, for example, scheduling maintenance tasks that need to be executed on different computers in a particular sequence.
Summary According to an aspect, there is provided a computer-implemented method for binding values to parameters including, via a processor: defining relationships between a plurality of tasks according to a dependency graph; maintaining in a memory a plurality of parameter contexts, each associated with a different scope of the tasks;
binding a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context; executing the first of the tasks, the first of the tasks using the value of the first parameter; determining a new value for assignment to the first parameter during execution of a second of the tasks; and storing the new value for the first parameter in the first parameter context; wherein the first parameter context is associated with the second of the tasks.

A further aspect provides a computer-readable storage medium encoded with computer-executable program instructions that when executed by one or more computers cause the one or more computers to perform operations comprising: defining a plurality of tasks according to a dependency graph; maintaining a plurality of parameter contexts, each associated with a different scope of the tasks; binding a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context;
executing the first of the tasks, the first of the tasks using the value of the first parameter;
determining a new value for assignment to the first parameter during execution of a second of the tasks; and storing the new value for the first parameter in the first parameter context;
wherein the first parameter context is associated with the second of the tasks.
There is also provided a computing system for binding values, the computing system including: a memory; and at least one processor configured to: define a plurality of tasks according to a dependency graph; maintain a plurality of parameter contexts in the memory, each associated with a different scope of the tasks; bind a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context;
execute the first of the tasks, the first of the tasks using the value of the first parameter;
determine a new value for assignment to the first parameter during execution of a second of the tasks; and store the new value for the first parameter in the first parameter context;
wherein the first parameter context is associated with the second of the tasks.
In accordance with a still further aspect, there is provided a task managing system for binding values including: a programmed processor means for defining a plurality of tasks according to a dependency graph; a programmed processor means for maintaining a plurality of parameter contexts, each associated with a different scope of the tasks; and a programmed processor means for binding a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context;
a programmed processor means for executing the first of the tasks, the first of the tasks using the value of the first parameter; a programmed processor means for determining a new value for assignment to
2 the first parameter during execution of a second of the tasks; and a programmed process means for storing the new value for the first parameter in the first parameter context; wherein the first parameter context is associated with the second of the tasks.
In a general aspect, the invention features a method for binding values of parameters. A number of tasks are defined according to a dependency graph.
Multiple parameter contexts are maintained, each associated with a different scope of the tasks. A
parameter used in a first of the tasks is bound to a value. This binding includes identifying a first of the contexts according to the dependency graph and retrieving the value for the parameter from the identified context.
This aspect can include one or more of the following features:
A value for assignment to a parameter is determined during execution of a second of the tasks. The value for the parameter is stored in the first parameter context, which is associated with the second of the tasks.
The dependency graph specifies an execution order for the tasks. Identifying the first of the contexts includes identifying tasks that necessarily execute prior to the first of the tasks. Identifying the first of the contexts also includes identifying tasks that assign a value to the first parameter and identifying a task that executes latest according to the dependency graph.
In another aspect, in general, the invention features a method for checking parameter access including accepting a specification of an ordering constraint for a plurality of tasks, in which at least some of the tasks are not ordered relative to one another. This checking includes, for each of one or more parameters, identifying the tasks that access those parameters, and checking for potential conflict in the access to the parameters by the identified tasks according to the specification of the ordering constraint.
Aspects of the invention can include one or more of the following advantages:
2a Use of explicit parameter passing between tasks enables checking for conflicts and ambiguities in parameter values. For example, such checking can ensure that parameters will take on the same values on repeated execution of a plan of tasks.
Scoping according to precedence in a dependency graph can be preferable to scoping according to nesting or plans and sub-plans because parameters can be passed between peer tasks in the plans.
Use of multiple parameter contexts can reduce name conflict by allowing the same parameter name to be used in different portions of a dependency graph subject to the multiple uses not introducing conflicts or ambiguities according to the scoping rules. This can allow different developers to author different portions of a plan without introducing name conflicts, and without having to explicitly identify different naming domains.
Other features and advantages are apparent from the following description, and from the claims.
Description of Drawings FIG. 1 is a task management system.
FIG. 2 is a dependency graph.
FIG. 3 is a diagram that includes a single global context in a task manager.
2b FIG. 4A is a diagram that includes multiple task-specific dynamic contexts in a task manager.
FIG. 4B is a diagram showing the relationships among the dynamic contexts according the dependency graph of FIG. 2.
FIG. 4C is a diagram showing scopes of tasks in the dependency graph of FIG.
2.
FIG. 5 is a task specification.
FIGS. 6A-B are examples of an application that uses dynamic parameters.
Description Overview Referring to FIG. 1, a task management system 100 controls the execution of tasks that make up an overall plan according to dependency constraints which specify which tasks in the plan must be completed before others of the tasks are executed. The dependency constraints are represented using a directed dependency graph 132 in which each node 134 of the graph corresponds to a different task in the plan and each directed arc 136 from one node to another indicates an ordering constraint such that the task corresponding to the first ("upstream") node must complete before the task corresponding to the second ("downstream") node is executed.
More generally, a task corresponding to a node can execute only after all its upstream tasks in the graph have completed. The dependency graph establishes a partial ordering of the tasks.
Therefore, tasks may be executed concurrently subject to constraints including the graph-based dependency constraints. Tasks may also be executed on different computers to achieve a parallelism speedup in overall execution of the plan.
Task management system 100 includes a task manager 110, which reads a plan specification 120. The plan specification includes a graph specification 122 defining the dependency graph 132, as well as a tasks specification 124 that defines the characteristics of each of the tasks. Graph specification 122 defines the dependency graph by a list of nodes 134, and a list of arcs 136 each identifying a source node and a destination node in the graph. Each node is associated with a task definition, which is specified in tasks specification 124, that
- 3-specifies the specific operations to be carried out in the task.
A task can be defined using a task template and a plan can be defined using a plan template. A template for a task or a plan can include references to parameters that are represented symbolically and are not necessarily bound to values until the task is executed. For example, a task template may include a reference to a parameter named FILENAME, and a reference to that parameter, represented using a "dollar-sign" syntax $F1LENAME, is bound to a specific value by the task manager 110 prior to or during execution of the task.
Some parameters are global to the entire plan and are assigned values when the entire plan is first executed. Such parameters can be formal parameters of the plan as a whole and may, for example, be specified in the command that invokes the overall plan or can be elicited interactively from a user who invokes the plan.
Other parameters do not necessarily have values specified when the plan is first executed. Such dynamic parameters may be assigned values during or as a result of execution of one or more tasks and then referenced by the templates of other tasks. For example, in FIG.
2, a parameter A can be set to a value by task 1, and then that value can be used by task 2.
Task management system 100 supports a number of additional features that are relevant to the discussion below. A first feature is that a node 134 may be associated with an entire "sub-plan" rather than a single task. A sub-plan is a lower-level plan of partially ordered tasks.
The sub-plan has a plan specification 120 of the same or similar structure as the top-level plan specification.
A second feature is that tasks have the ability to modify plan specification 120 through their execution. In particular, a task has the ability to add additional nodes 134 and arcs 136 to graph 132, and define characteristics for the tasks associated with the added nodes using services provided by the task manager 110. For example, a first task can generate a sub-plan using a template for a plan having unknown (before runtime) number of tasks to perform. The number of tasks in the template is bound at runtime using a dynamic parameter.
The generated sub-plan is incorporated into the higher-level plan of the first task by the task manager 110.
-4-Binding Approaches Task manager 110 uses one of a number of alternative approaches to parameter binding as described below. Alternatively, the task manager 110 uses a combination of the binding approaches, for example, using different approaches for different parameters according to declarations of those parameters. One aspect in which some (or all) of the approaches differ is in the scope of parameter definitions.
Approach 1 Referring to FIG. 3, in a first binding approach, task manager 110 maintains a single global context 310 for all parameters. The task manager 110 maintains a single copy of any parameter. That is, it maintains a data structure 320 that associates each of a number of different parameter names 322 with a current value 324 of that parameter (where a value 324 can be empty if the corresponding parameter has not yet been assigned a value). This same data structure 320 can be used to hold values of global parameters as well as those dynamically assigned by tasks as they execute.
Referring to FIGS. 2 and 3, in an example of this first approach, a task 1 assigns a value FOO to parameter A, as indicated by the assignment A=F00. Task manager 110 receives the assigned value, FOO, from task 1 (indicated by the arrow from task 1 to the value cell 324 for parameter A) and stores the value in a value cell 324. The task manager 110 later provides the value FOO to task 2 in order to bind a reference to parameter A in the template for task 2. In this example, task 2 includes an assignment B=SA/BAZ, which can result in the parameter B
being assigned the value FOO/BAZ. The task manager 110 receives the assigned value for B
which it stores in global context 310. According to graph 132, task 2 executes only after task 1 is completed. Therefore, the assignment of FOO to the parameter A is guaranteed to have been performed prior to the execution of task 2.
In general, groups of multiple tasks of a plan can be executed concurrently.
For a particular dependency graph, two or more tasks that are not ordered relative to one another could potentially be executed concurrently or in an unpredictable sequence if there is no direct path between them in the dependency graph. In FIG. 2, tasks 2 and 3 are such unordered tasks.
In the example in FIG. 2, task 3 assigns a value of BAR to parameter A. If task 3 executes
- 5-
6 before task 2, then the parameter B is assigned a value of BAR/BAZ in task 2, while if task 3 executes after task 2, the parameter B is assigned a value of FOO/BAZ.
In a number of applications, the uncertainty of which value will be assigned to B is undesirable. One approach to preventing such a situation is to perform a static checking of a dependency graph and prohibiting use of a plan that exhibits such uncertainty.
The check includes identifying groups of tasks that are not ordered by the graph. If a particular parameter is used by one task in the group and assigned by another task in the group, a potential ambiguity occurs. Similarly, if two tasks in such a group assign values to a particular parameter, there is also a conflict. In general, in any group of unordered tasks in which multiple tasks access a particular parameter no task may assign a value to that parameter. This static check would identify the conflict between tasks 2 and 3 with respect to their reference to parameter A. The task management system 100 handles conflicts identified by the static check (e.g., using error handling mechanisms). Note that in this example, tasks 2 and 5 can also execute in an unpredictable order, but both use rather than assign a value of parameter A and therefore do not represent a conflict or ambiguity.
The static checking of the dependency graph is performed by the task manager before execution of the plan or alternatively can be performed in a pre-process by another module that validates the plan specification. In the situation in which the graph is modified by execution of a task, the static check is performed again or augmented to account for the modification of the graph.
For plans that include sub-plans, two variants of this approach can be used.
In a first variant, the task manager 110 maintains a single context for all parameters for all levels of sub-plans. In static checking of a plan that includes a node with a sub-plan, an assignment of a parameter by any task in that sub-plan conflicts with any unordered task (with respect to the node with that sub-plan) in the higher-level plan that uses the value of that parameter.
In a second variant involving sub-plans, the task manager 110 maintains a separate context for parameters of each sub-plan, and the specification of a sub-plan explicitly identifies which parameters are exported from the sub-plan to the next higher-level plan when the sub-plan is completed. In such a variant, the static checking involves conflict between an exported parameter of a sub-plan and access to that parameter in a task in the next higher-level plan that is not ordered with execution of the sub-plan.
In another version of this approach, the task manager 110 prohibits repeated assignment of values to any particular global dynamic parameter.
Approach 2 Referring to FIG. 4A, in a second approach task manager 110 maintains a global context 310, for parameters that are set for the plan as a whole, and multiple dynamic contexts 410, each associated with a different node of the dependency graph. Each task that assigns a parameter generates a new "instance" of that parameter and stores (via the task manager 110) the assigned value in the dynamic context 410 associated with the node corresponding to that task. The task manager 110 can therefore store multiple values associated with a parameter, each value associated with a different instance of the parameter. For any particular parameter (e.g., A), the task manager 110 maintains a separate value of the parameter associated with each task in which that parameter is assigned (since each task corresponds to a separate node and a separate dynamic context 410).
Referring to FIGS. 4B and 4C, each instance of a parameter and its associated dynamic context 410 has a scope of tasks that determines which tasks "see" the value assigned to that instance. Each dynamic context 410 is associated with a different scope of tasks based on relationships among the dynamic contexts 410 following a pattern 420 according to the dependency graph 132. For example, the instance of parameter A that is assigned A=F00 in task 1 (stored in a dynamic context 410) has a scope of task 430 that is different from the scope of tasks 432 associated with the instance of parameter A that is assigned A=BAR in task 3.
The scope of tasks 434 associated with the instance of parameter B that is assigned B=SA/BAZ
in task 2 includes task 4 but not task 5. In this approach, the scope of tasks or "scope" of an instance of a parameter (and its associated dynamic context 410) includes only the task that assigned the value to the instance of the parameter and tasks that are ordered after ("downstream of') that task. When a task references a parameter, the task manager 110 binds to that reference a value of an instance of the parameter whose scope includes the task.
When task 1 assigns a value FOO to parameter A, the task manager 110 stores the value in a dynamic context 410 associated with task 1. Similarly, when task 3 assigns a value BAR
- 7-to parameter A, the task manager 110 stores the value in a dynamic context 410 associated with task 3. Task 2 is within the scope of the instance of parameter A which is assigned a value A----F00 in task 1, but is not in the scope of the instance of parameter A
that is assigned A=BAR in task 3. Therefore, under this scoping approach, there is no uncertainty or ambiguity regarding the value of B in task 2. The task manager 110 binds the reference to parameter A in task 2 with the value from the dynamic context 410 for task 1. Task 2 assigns B the valve FOO/BAZ, which it provides to the task manager 110 for storage in the dynamic context 410 associated with task 2.
In general, during execution the task manager 110 binds a value to a reference to a parameter in a task by considering upstream paths (i.e., direct paths through a single arc or indirect paths through already executed tasks) through the graph that terminate with a task that assigns a value to the parameter. In the example, when binding the parameter A
in task 2, the task manager 110 follows the upstream path to task 1, where the parameter has been set.
In the example in FIG. 2, task 4 is potentially within the scope of both the instance of parameter A of task 1 and that of task 3. That is, there is an upstream path from task 4 to both task 1 (indirectly through task 2) and to task 3 (directly) each of which assigns a value to A.
However, in this example, there is no ambiguity because task 3 is constrained to execute after task 1. If there are multiple upstream assignments of a parameters and if there is one task that is ordered after all the others, then there is no ambiguity and the value assigned to the instance of the parameter in the dynamic context 410 of that last task is used.
As with the first approach, potential ambiguity can be checked statically from the dependency graph. Specifically, for each reference to a parameter at a node (task) in the graph, the upstream tasks that could assign values to the parameter are first identified. If there is no such task, then the parameter would have an undefined value or is provided from a global context for the plan. If there is exactly one upstream task, then there is no ambiguity. If there are multiple upstream tasks that assign values to the parameter there is an ambiguity unless there is a single one of those tasks that is ordered after the others (i.e., constrained to be executed after the other by the dependency graph).
- 8-Implementation approaches Referring to FIG. 5, a task specification 500 (which is contained in tasks specification 124 in FIG. 1) includes a declaration section 510 and an instruction section 520. The declaration section 510 identifies (e.g., using a declaration statement or statements) the dynamic parameters that are assigned through execution of the corresponding task. The task specification shown in FIG. 5 corresponds to task 1 in FIG. 2. The parameter A
is declared as being dynamically assigned by the task, illustrated as the declaration "dynamic A" 512. The declaration section 510 also identifies references to parameters that are required to be bound by the task manager 110 according to the scopes of the available parameters.
The instruction section 520 includes computer instruction (e.g., programming language or other procedural statements) for the operations to be carried out by the task, as well as instructions for assigning values to dynamic parameters, such as the instruction "A=F00" 522.
Such computer instruction, for example, can be specified by a user or automatically generated by a computer process.
Task manager 110 is able to process the declaration sections of the task specifications for a plan to determine which parameters are to be assigned, or referenced, or both during execution of the corresponding tasks. The task manager 110 can distinguish between an assignment or a reference by additionally processing the instruction sections of the task specifications, or alternatively, by the use of a separate type of declaration statement for dynamic parameter assignment or reference. For example, for the example shown in FIG. 2, the task manager 110 can parse the task specifications to determine that task 2 references parameter A while task 3 assigns a value to A. This allows the task manager 110 to perform a static check, such as a static check for the first approach (see FIG. 3) in which the reference in task 2 and the assignment in task 3 would conflict.
When the declaration section identifies the parameters that are to be referenced, the task manager 110 is able to assemble an environment of parameters and their bound values prior to executing the instructions of instruction section 520. Having assembled the environment for a task, the task manager 110 invokes a process that executes the instructions of the instruction section of the task specification. That process then accesses the parameter values from the environment that has been created for it. During execution, processing of an assignment
- 9-instruction for a dynamic parameter results in a (parameter,value) pair being recorded in an output of the task execution, a dynamic parameter assignment section 530. For example, the pair (A,F00) 532 would be recorded in the assignment section 530. The task manager 110 receives the assignment section 530 after execution of the instructions of the instruction section 520, and uses its contents to update the parameter values.
In one specific implementation of this approach, the instruction section includes statements in a shell scripting language such as "ksh." Prior to invoking a shell process to interpret the instructions, the task manager 110 uses the declaration section 510 to assemble an environment of parameter values that will be used in processing the shell instructions. In this implementation, assignment statements take the form of output statements to the text output of the script. That is, rather than using an assignment "A=F00", the scripting instructions would include a statement such as "PRINT A=F00", resulting in a line "A=F00" being output. The text output of the script forms the dynamic parameter assignment section 530.
The task manager 110 receives the text output, and interprets the assignment statements storing the values in the appropriate dynamic contexts 410.
In another implementation, the process interpreting the instructions in instruction section 520 is in communication with the task manager 110, and the task manager 110 services requests to store or to access values of dynamic parameters. When a parameter reference needs to be bound to a value to execute an instruction, the process sends a request to the task manager 110, which determines the appropriate dynamic context 410 for that parameter and returns the value in that dynamic context 410. Similarly, when a parameter is assigned a value, the process sends a request to store the value in the appropriate dynamic context 410.
In an alternative implementation, task specification 500 declaration section 510 does not identify the parameters that will be needed by the task during execution, and a suitable communication mechanism is not available to request the parameter values while the instructions of the task are being executed. In this implementation, the task manager 110 assembles a complete list of all parameter instances whose scopes include the task. For parameters for which binding would cause a potential ambiguity (for example because the parameter was assigned a value by multiple upstream unordered tasks), the parameter value is replace with an indicator, such as "ambiguous_value", so that an error would occur should that
- 10-parameter be referenced in the task.
When the declaration sections 510 do not include declarations of the parameters that will be dynamically assigned and/or referenced in each task, static checking of a dependency graph may not be possible. However, the task manager 110 is still able to identify some potential conflicts that may occur during execution of a plan. For example, using the scoping approach with a single global context (see FIG. 3), if the task manager 110 detects during execution of a plan that multiple unordered tasks have assigned values to a parameter, the task manager 110 can identify the potential conflict. The task management system 100 can then handle conflicts identified by the task manager.
Examples Referring to FIG. 6A, an example of use of dynamically assigned parameters involves the processing of a number of data files 641-643, each by a different task 631-633 in an overall plan. The file names of the data files 641-643 are not known when the plan is first executed.
Rather, the file names are listed in a text file 622, the name of which is provided as a global parameter LIST_FILE for the overall plan.
The plan includes a first task 620, which is responsible for opening the LIST_FILE file 622, and reading its contents. This task 620 opens the file and reads the contents, determining the number of separate data files listed. In this example, the list file 622 has three file names, FNAMEl.DAT, FNAME2.DAT, and FNAME3.DAT. The task retrieves the listed files 643, for example, using a file transfer protocol across a data network. Task 1 assigns values to a number of dynamic parameters, NUM_INPUT, and DATA_FILE _1 through DATA_FILE_3, and passes this assignment information to the task manager 110 which records it in an appropriate context 610 (e.g., a global context, or a dynamic context associated with task 1, depending on the scoping approach used).
In this first version of the example, the initially specified plan has exactly three instances of tasks for processing the input files, tasks 2-4 (631-633). Each task retrieves the name of its corresponding input data file by referencing a dynamic parameter assigned by task 1.
- 11-A variant of this example takes advantage of tasks being able to modify the plan specification. The plan initially has only a single task, the GET NAMES task 1 (620). After the GET NAMES task has determined that there are three data'. files to process, the task makes use of services of the task manager 110 to create the three processing tasks 631-633 and connect them to itself (task 1). Then after task 1 completes, the task manager 110 executes the newly-created tasks 631-633 as described above.
Alternatively, in another variant of this example, the task manager 110 can use a plan template having a modifiable number of tasks. Referring to FIG. 6B, the plan initially has two tasks, the GET NAMES task 1 (620), and a PROCESS task 2 (650). After the GET
NAMES
task has determined that there are three data files to process, the PROCESS
task creates a sub-plan using a plan template having a modifiable number of tasks. The PROCESS
task references the dynamic parameter NUM_INPUT to create a sub-plan with three (unordered) processing tasks 631-633. The task manager 110 executes the tasks 631-633 in the newly-created sub-plan as described above.
The parameter scoping approach described above can be implemented using software for execution on a computer. For instance, the software forms procedures in one or more computer programs that execute on one or more programmed or programmable computer systems (which may be of various architectures such as distributed, client/server, or grid) each including at least one processor, at least one data storage system (including volatile and non-volatile memory and/or storage elements), at least one input device or port, and at least one output device or port. The software may form one or more modules of a larger program, for example, that provides other services related to the design and configuration of computation graphs.
The software may be provided on a medium, such as a CD-ROM, readable by a general or special purpose programmable computer or delivered (encoded in a propagated signal) over a network to the computer where it is executed. All of the functions may be performed on a special purpose computer, or using special-purpose hardware, such as coprocessors. The software may be implemented in a distributed manner in which different parts of the computation specified by the software are performed by different computers.
Each such computer program is preferably stored on or downloaded to a storage media or device (e.g.,
- 12-solid state memory or media, or magnetic or optical media) readable by a general or special purpose programmable computer, for configuring and operating the computer when the storage media or device is read by the computer system to perform the procedures described herein.
The inventive system may also be considered to be implemented as a computer-readable storage medium, configured with a computer program, where the storage medium so configured causes a computer system to operate in a specific and predefined manner to perform the functions described herein.
It is to be understood that the foregoing description is intended to illustrate and not to limit the scope of the invention, which is defined by the scope of the appended claims. Other embodiments are within the scope of the following claims.
- 13-

Claims (16)

CLAIMS:
1. A computer-implemented method for binding values to parameters including, via a processor:
defining relationships between a plurality of tasks according to a dependency graph;
maintaining in a memory a plurality of parameter contexts, each associated with a different scope of the tasks;
binding a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context;
executing the first of the tasks, the first of the tasks using the value of the first parameter;
determining a new value for assignment to the first parameter during execution of a second of the tasks; and storing the new value for the first parameter in the first parameter context;
wherein the first parameter context is associated with the second of the tasks.
2. The method of claim I wherein the dependency graph specifies an execution order for the tasks, and wherein identifying the first of the contexts includes identifying tasks that necessarily execute prior to the first of the tasks.
3. The method of claim 2 wherein identifying the first of the contexts includes identifying tasks that assign a value to the first parameter.
4. The method of claim 3 wherein identifying tasks that assign a value to the first parameter includes identifying a task that executes latest according to the dependency graph.
5. A computer-readable storage medium encoded with computer-executable program instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:
defining a plurality of tasks according to a dependency graph;
maintaining a plurality of parameter contexts, each associated with a different scope of the tasks;
binding a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context;
executing the first of the tasks, the first of the tasks using the value of the first parameter;
determining a new value for assignment to the first parameter during execution of a second of the tasks; and storing the new value for the first parameter in the first parameter context;
wherein the first parameter context is associated with the second of the tasks.
6. The medium of claim 5 wherein the dependency graph specifies an execution order for the tasks, and wherein identifying the first of the contexts includes identifying tasks that necessarily execute prior to the first of the tasks.
7. The medium of claim 6 wherein identifying the first of the contexts includes identifying tasks that assign a value to the first parameter.
8. The medium of claim 7 wherein identifying tasks that assign a value to the first parameter includes identifying a task that executes latest according to the dependency graph.
9. A computing system for binding values, the computing system including:
a memory; and at least one processor configured to:
define a plurality of tasks according to a dependency graph;
maintain a plurality of parameter contexts in the memory, each associated with a different scope of the tasks;
bind a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context;
execute the first of the tasks, the first of the tasks using the value of the first parameter;
determine a new value for assignment to the first parameter during execution of a second of the tasks; and store the new value for the first parameter in the first parameter context;
wherein the first parameter context is associated with the second of the tasks.
10. The system of claim 9 wherein the dependency graph specifies an execution order for the tasks, and wherein identifying the first of the contexts includes identifying tasks that necessarily execute prior to the first of the tasks.
11. The system of claim 10 wherein identifying the first of the contexts includes identifying tasks that assign a value to the first parameter.
12. The system of claim 11 wherein identifying tasks that assign a value to the first parameter includes identifying a task that executes latest according to the dependency graph.
13. A task managing system for binding values including:
a programmed processor means for defining a plurality of tasks according to a dependency graph;

a programmed processor means for maintaining a plurality of parameter contexts, each associated with a different scope of the tasks; and a programmed processor means for binding a value to a first parameter used in a first of the tasks, including identifying a first of the parameter contexts according to the dependency graph and retrieving the value for the parameter from the identified context;
a programmed processor means for executing the first of the tasks, the first of the tasks using the value of the first parameter;
a programmed processor means for determining a new value for assignment to the first parameter during execution of a second of the tasks; and a programmed process means for storing the new value for the first parameter in the first parameter context;
wherein the first parameter context is associated with the second of the tasks.
14. The system of claim 13 wherein the dependency graph specifies an execution order for the tasks, and wherein identifying the first of the contexts includes identifying tasks that necessarily execute prior to the first of the tasks.
15. The system of claim 14 wherein identifying the first of the contexts includes identifying tasks that assign a value to the first parameter.
16. The system of claim 15 wherein identifying tasks that assign a value to the first parameter includes identifying a task that executes latest according to the dependency graph.
CA2558826A 2004-03-08 2005-03-08 Dependency graph parameter scoping Active CA2558826C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CA2880884A CA2880884C (en) 2004-03-08 2005-03-08 Dependency graph parameter scoping

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US10/795,374 US7594227B2 (en) 2004-03-08 2004-03-08 Dependency graph parameter scoping
US10/795,374 2004-03-08
PCT/US2005/007923 WO2005086906A2 (en) 2004-03-08 2005-03-08 Dependency graph parameter scoping

Related Child Applications (1)

Application Number Title Priority Date Filing Date
CA2880884A Division CA2880884C (en) 2004-03-08 2005-03-08 Dependency graph parameter scoping

Publications (2)

Publication Number Publication Date
CA2558826A1 CA2558826A1 (en) 2005-09-22
CA2558826C true CA2558826C (en) 2015-05-12

Family

ID=34976229

Family Applications (2)

Application Number Title Priority Date Filing Date
CA2880884A Active CA2880884C (en) 2004-03-08 2005-03-08 Dependency graph parameter scoping
CA2558826A Active CA2558826C (en) 2004-03-08 2005-03-08 Dependency graph parameter scoping

Family Applications Before (1)

Application Number Title Priority Date Filing Date
CA2880884A Active CA2880884C (en) 2004-03-08 2005-03-08 Dependency graph parameter scoping

Country Status (8)

Country Link
US (2) US7594227B2 (en)
EP (2) EP1735704A4 (en)
JP (2) JP4891225B2 (en)
CN (2) CN100426239C (en)
AU (1) AU2005221142B2 (en)
CA (2) CA2880884C (en)
EA (1) EA012057B1 (en)
WO (1) WO2005086906A2 (en)

Families Citing this family (59)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7594227B2 (en) * 2004-03-08 2009-09-22 Ab Initio Technology Llc Dependency graph parameter scoping
US7650347B2 (en) * 2004-10-01 2010-01-19 Computer Associates Think, Inc. System and method for job scheduling and distributing job scheduling
US8171474B2 (en) 2004-10-01 2012-05-01 Serguei Mankovski System and method for managing, scheduling, controlling and monitoring execution of jobs by a job scheduler utilizing a publish/subscription interface
US7756801B2 (en) * 2004-11-15 2010-07-13 Palo Alto Research Center Incorporated Systems and methods for architecture independent programming and synthesis of network applications
US7877748B2 (en) * 2004-11-19 2011-01-25 The United States Of America As Represented By The Secretary Of The Air Force Method and apparatus for timing information flow in a distributed system
US7877350B2 (en) 2005-06-27 2011-01-25 Ab Initio Technology Llc Managing metadata for graph-based computations
US20070006070A1 (en) * 2005-06-30 2007-01-04 International Business Machines Corporation Joining units of work based on complexity metrics
WO2007085121A1 (en) * 2006-01-26 2007-08-02 Intel Corporation Scheduling multithreaded programming instructions based on dependency graph
US20080040725A1 (en) * 2006-08-11 2008-02-14 Barrie Jon Moss Method and apparatus for a parallel model of tasks including abstracted execution and software development
US8037466B2 (en) 2006-12-29 2011-10-11 Intel Corporation Method and apparatus for merging critical sections
US8225326B2 (en) * 2008-03-17 2012-07-17 Intel Corporation Future scheduling by direct representation of possible dependencies
US8276165B2 (en) * 2008-05-29 2012-09-25 Microsoft Corporation Continuation-based runtime callback invocation
EP2370873B1 (en) 2008-12-02 2021-02-03 Ab Initio Technology LLC Visualizing relationships between data elements
US8266477B2 (en) 2009-01-09 2012-09-11 Ca, Inc. System and method for modifying execution of scripts for a job scheduler using deontic logic
WO2010093879A1 (en) 2009-02-13 2010-08-19 Ab Initio Technology Llc Managing task execution
US8683432B2 (en) * 2009-03-20 2014-03-25 Microsoft Corporation Providing execution context in continuation based runtimes
US20100293538A1 (en) * 2009-05-15 2010-11-18 Microsoft Corporation Dynamic program updating in a continuation based runtime
CN103069385B (en) * 2010-06-15 2016-12-28 起元技术有限责任公司 System and method for dynamic load calculating based on figure
US8874457B2 (en) * 2010-11-17 2014-10-28 International Business Machines Corporation Concurrent scheduling of plan operations in a virtualized computing environment
US8935705B2 (en) * 2011-05-13 2015-01-13 Benefitfocus.Com, Inc. Execution of highly concurrent processing tasks based on the updated dependency data structure at run-time
US8595743B2 (en) * 2012-05-01 2013-11-26 Concurix Corporation Network aware process scheduling
US8726255B2 (en) 2012-05-01 2014-05-13 Concurix Corporation Recompiling with generic to specific replacement
US20120222043A1 (en) * 2012-05-01 2012-08-30 Concurix Corporation Process Scheduling Using Scheduling Graph to Minimize Managed Elements
US8650538B2 (en) 2012-05-01 2014-02-11 Concurix Corporation Meta garbage collection for functional code
US9417935B2 (en) * 2012-05-01 2016-08-16 Microsoft Technology Licensing, Llc Many-core process scheduling to maximize cache usage
US9047196B2 (en) 2012-06-19 2015-06-02 Concurix Corporation Usage aware NUMA process scheduling
US8700838B2 (en) 2012-06-19 2014-04-15 Concurix Corporation Allocating heaps in NUMA systems
FI20125700A (en) * 2012-06-21 2013-12-22 Tekla Corp Shared data with relationship information
US8793669B2 (en) 2012-07-17 2014-07-29 Concurix Corporation Pattern extraction from executable code in message passing environments
US8707326B2 (en) 2012-07-17 2014-04-22 Concurix Corporation Pattern matching process scheduler in message passing environment
US9575813B2 (en) 2012-07-17 2017-02-21 Microsoft Technology Licensing, Llc Pattern matching process scheduler with upstream optimization
US9043788B2 (en) 2012-08-10 2015-05-26 Concurix Corporation Experiment manager for manycore systems
FR2995106B1 (en) * 2012-09-03 2015-08-14 Bull Sas METHOD AND DEVICE FOR PROCESSING CONTROLS IN A SET OF COMPUTER ELEMENTS
US9852153B2 (en) 2012-09-28 2017-12-26 Ab Initio Technology Llc Graphically representing programming attributes
US10423424B2 (en) * 2012-09-28 2019-09-24 Nvidia Corporation Replicated stateless copy engine
US8656135B2 (en) 2012-11-08 2014-02-18 Concurix Corporation Optimized memory configuration deployed prior to execution
US8656134B2 (en) 2012-11-08 2014-02-18 Concurix Corporation Optimized memory configuration deployed on executing code
US8607018B2 (en) 2012-11-08 2013-12-10 Concurix Corporation Memory usage configuration based on observations
US9507682B2 (en) 2012-11-16 2016-11-29 Ab Initio Technology Llc Dynamic graph performance monitoring
US10108521B2 (en) 2012-11-16 2018-10-23 Ab Initio Technology Llc Dynamic component performance monitoring
US9116603B2 (en) 2012-12-26 2015-08-25 Ab Initio Technology Llc Managing interactions with data having membership in multiple groupings
US9274926B2 (en) 2013-01-03 2016-03-01 Ab Initio Technology Llc Configurable testing of computer programs
US9436589B2 (en) 2013-03-15 2016-09-06 Microsoft Technology Licensing, Llc Increasing performance at runtime from trace data
US9207978B2 (en) * 2013-10-09 2015-12-08 Wipro Limited Method and system for efficient execution of ordered and unordered tasks in multi-threaded and networked computing
CA2932763C (en) 2013-12-05 2022-07-12 Ab Initio Technology Llc Managing interfaces for dataflow graphs composed of sub-graphs
US20160034258A1 (en) * 2014-07-31 2016-02-04 Lexmark International Technology S.A. System and Methods of Generating Build Dependencies Radiator
WO2016186651A1 (en) * 2015-05-19 2016-11-24 Hewlett Packard Enterprise Development Lp Updating inference graph absent node locking
US10657134B2 (en) 2015-08-05 2020-05-19 Ab Initio Technology Llc Selecting queries for execution on a stream of real-time data
WO2017090005A1 (en) * 2015-11-27 2017-06-01 Cadens Imagerie Medicale Inc. Method and system for executing a function for processing data using a server
WO2017095384A1 (en) * 2015-11-30 2017-06-08 Hewlett-Packard Enterprise Development LP Task-based conflict resolution
CN108475189B (en) 2015-12-21 2021-07-09 起元技术有限责任公司 Method, system and computer readable medium for sub-graph interface generation
US10228973B2 (en) * 2016-03-08 2019-03-12 Hulu, LLC Kernel policy optimization for computing workloads
US10769113B2 (en) 2016-03-25 2020-09-08 Microsoft Technology Licensing, Llc Attribute-based dependency identification for operation ordering
US10152349B1 (en) * 2016-09-27 2018-12-11 Juniper Networks, Inc. Kernel scheduling based on precedence constraints and/or artificial intelligence techniques
WO2018102691A1 (en) 2016-12-01 2018-06-07 Ab Initio Technology Llc Generating, accessing, and displaying lineage metadata
US10802945B2 (en) * 2016-12-07 2020-10-13 Ab Initio Technology Llc Differencing of executable dataflow graphs
US10552218B2 (en) * 2017-01-06 2020-02-04 Microsoft Technology Licensing, Llc Dynamic context of tasks
JP7267819B2 (en) * 2019-04-11 2023-05-02 株式会社 日立産業制御ソリューションズ Parallel task scheduling method
CN110471754B (en) * 2019-08-22 2022-04-05 中国工商银行股份有限公司 Data display method, device, equipment and storage medium in job scheduling

Family Cites Families (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4893234A (en) * 1987-01-15 1990-01-09 United States Department Of Energy Multi-processor including data flow accelerator module
US5617510A (en) * 1989-06-13 1997-04-01 Schlumberger Technology Corporation Dependency graph solution for constraint systems
IL93239A (en) * 1990-02-01 1993-03-15 Technion Res & Dev Foundation High flow-rate synchronizer/schedular apparatus for multiprocessors
US6185516B1 (en) * 1990-03-06 2001-02-06 Lucent Technologies Inc. Automata-theoretic verification of systems
WO1993020511A1 (en) * 1992-03-31 1993-10-14 Aggregate Computing, Inc. An integrated remote execution system for a heterogenous computer network environment
US5689711A (en) * 1995-04-21 1997-11-18 Bardasz; Theodore Method and apparatus for representing data dependencies in software modeling systems
US5808625A (en) * 1995-08-04 1998-09-15 Silicon Graphics Incorporated System and method for using dependency graphs for the control of a graphics creation process
AU735024B2 (en) * 1997-07-25 2001-06-28 British Telecommunications Public Limited Company Scheduler for a software system
US6675189B2 (en) * 1998-05-28 2004-01-06 Hewlett-Packard Development Company, L.P. System for learning and applying integrated task and data parallel strategies in dynamic applications
US6480876B2 (en) * 1998-05-28 2002-11-12 Compaq Information Technologies Group, L.P. System for integrating task and data parallelism in dynamic applications
US6490611B1 (en) * 1999-01-28 2002-12-03 Mitsubishi Electric Research Laboratories, Inc. User level scheduling of inter-communicating real-time tasks
US20030005407A1 (en) * 2000-06-23 2003-01-02 Hines Kenneth J. System and method for coordination-centric design of software systems
US7164422B1 (en) * 2000-07-28 2007-01-16 Ab Initio Software Corporation Parameterized graphs with conditional components
US6574788B1 (en) * 2000-11-13 2003-06-03 Reshape, Inc. Method and system for automatically generating low level program commands as dependency graphs from high level physical design stages
US7155720B2 (en) * 2001-10-26 2006-12-26 Hewlett-Packard Development Company, L.P. Dynamic task assignment in workflows
US7191329B2 (en) * 2003-03-05 2007-03-13 Sun Microsystems, Inc. Automated resource management using perceptron prediction
US7152157B2 (en) * 2003-03-05 2006-12-19 Sun Microsystems, Inc. System and method for dynamic resource configuration using a dependency graph
US7594227B2 (en) * 2004-03-08 2009-09-22 Ab Initio Technology Llc Dependency graph parameter scoping

Also Published As

Publication number Publication date
CA2880884C (en) 2016-12-06
EP2541411A2 (en) 2013-01-02
CN101276290A (en) 2008-10-01
AU2005221142A1 (en) 2005-09-22
JP5897816B2 (en) 2016-03-30
US7594227B2 (en) 2009-09-22
CA2558826A1 (en) 2005-09-22
EP2541411A3 (en) 2014-02-12
EP1735704A4 (en) 2014-02-12
CA2880884A1 (en) 2005-09-22
US20050257221A1 (en) 2005-11-17
WO2005086906A2 (en) 2005-09-22
EA200601657A1 (en) 2007-02-27
CN1965296A (en) 2007-05-16
AU2005221142B2 (en) 2011-05-12
US8082544B2 (en) 2011-12-20
US20090089552A1 (en) 2009-04-02
CN101276290B (en) 2015-04-01
CN100426239C (en) 2008-10-15
EA012057B1 (en) 2009-08-28
JP4891225B2 (en) 2012-03-07
WO2005086906A3 (en) 2006-11-16
JP2007528081A (en) 2007-10-04
EP1735704A2 (en) 2006-12-27
JP2011170879A (en) 2011-09-01

Similar Documents

Publication Publication Date Title
CA2558826C (en) Dependency graph parameter scoping
JP5851396B2 (en) Processing method
US20080320453A1 (en) Type inference and late binding
Danelutto et al. Data stream processing via code annotations
Kale et al. Toward a standard interface for user-defined scheduling in OpenMP
US20220121498A1 (en) Combination of multiple data processing and machine learning frameworks for a target hardware
Guyer et al. Optimizing the use of high performance software libraries
Evans et al. Circuitflow: A domain specific language for dataflow programming
Thoman et al. The celerity high-level api: C++ 20 for accelerator clusters
Chevalier et al. Sharing ghost variables in a collection of abstract domains
AU2011204916B2 (en) Managing parameter access in a task managing system
US11144290B2 (en) Method and apparatus for enabling autonomous acceleration of dataflow AI applications
CN105393216B (en) Run-time memory is adjusted
Henrio et al. Active objects with deterministic behaviour
JP3241214B2 (en) Distributed processing apparatus and process execution method
Albert et al. A formal, resource consumption-preserving translation of actors to haskell
Gerasimov The DSL for composing functions for FaaS platform
Olsson et al. RJ: a Java package providing JR‐like concurrent programming
Chow Redundancy Elimination
Voss et al. Jumping Right In:“Hello, TBB!”
Mohaqeqi et al. An executable semantics for synchronous task graphs: From sdrt to ada
Mantione et al. Hashed SSA Form: HSSA
Wozniak Swift/T Guide
Sakr et al. Advanced Giraph Programming
Lorenzen et al. Eine abstrakte Maschine für eine nebenläufige (und parallele) Constraint-funktionale Programmiersprache

Legal Events

Date Code Title Description
EEER Examination request