Method and system for scheduling tasks in a computing system
In general, embodiments of the invention relate to a method and computing system for scheduling tasks (functions or routines) dynamically from Input/Output (I/O) operations that may be received from a client. The scheduling or ordering of the tasks play an important role in the overall latency of the execution of IO operations, as each task may consume significant amount of computing resources.
1. A method for scheduling tasks, the method comprising:
receiving, by a computing system, an Input/Output (I/O) operation from a client;
identifying a plurality of tasks associated with the I/O operation;
identifying dependencies for each of the plurality of tasks;
creating a directed graph of the tasks using the identified dependencies;
assigning a first set of edge weights to each task of the plurality of tasks based on the identified dependencies, wherein at least one edge weight of the first set of edge weights is non-zero;
ordering the tasks based on the first set of edge weights to obtain a task order;
selecting a highest order task based on the task order, wherein the highest ordered task does not have a zero-edge weight;
executing the highest ordered task;
updating, dynamically, based on an execution of the highest order task, the first set of edge weights for remaining tasks to obtain an updated task order;
selecting a highest order task from the updated task order;
executing the highest ordered task from the updated task order, wherein tasks with a zero-edge weight are not executed; and
updating, dynamically, based on an execution of the highest order task from the updated task order, a second set of edge weights for remaining tasks in updated task order to obtain a second updated task order, wherein at least one of the first set of edge weights is modified after the execution of the highest ordered task, and wherein the task dependencies are generated by a compiler during compile-time.
2. The method of claim 1 ,
wherein the task dependencies are specified using a directed graph comprising a plurality of edges; and
wherein each of a first set of edges in the directed graph is associated with one of the first set of edge weights.
3. The method of claim 1 , wherein each at least portion of the task dependencies are specified using a directed graph.
4. The method of claim 3 , wherein the task order is determined using a depth-first search, performed on the directed graph.
5. A non-transitory computer readable medium (CRM) storing instructions for scheduling tasks, the instructions comprising functionality for:
receiving, by a computing system, an Input/Output (I/O) operation from a client;
identifying a plurality of tasks associated with the I/O operation;
identifying dependencies for each of the plurality of tasks;
creating a directed graph of the tasks using the identified dependencies;
assigning a first set of edge weights to each task of the plurality of tasks based on the identified task dependencies, wherein at least one edge weight of the first set of edge weights is non-zero;
ordering the tasks based on the first set of edge weights to obtain a task order;
selecting a highest order task based on the task order, wherein the highest ordered task does not have a zero-edge weight;
executing the highest ordered task;
updating, dynamically, based on an execution of the highest order task, the first set of edge weights for remaining tasks to obtain an updated task order;
selecting a highest order task from the updated task order;
executing the highest ordered task from the updated task order, wherein tasks with a zero-edge weight are not executed; and
updating, dynamically, based on an execution of the highest order task from the updated task order, a second set of edge weights for remaining tasks in updated task order to obtain a second updated task order, wherein at least one of the first set of edge weights is modified after the execution of the highest ordered task, and wherein the task dependencies are generated by a compiler during compile-time.
6. The CRM of claim 5 , wherein the task dependencies are specified using a directed graph comprising a plurality of edges; and wherein each of a first set of edges in the directed graph is associated with one of the first set of edge weights.
7. The CRM of claim 5 , wherein each of the task dependencies are specified using a directed graph.
8. The CRM of claim 7 , wherein the task order is determined using a depth-first search, performed on the directed graph.
9. A computing system, comprising:
a processor;
a scheduler; and
wherein the scheduler, when executed by the processor, enables the scheduler to perform a method, the method comprising:
receiving, by a computing system, an Input/Output (I/O) operation from a client;
identifying a plurality of tasks associated with the I/O operation;
identifying dependencies for each of the plurality of tasks;
creating a directed graph of the tasks using the identified dependencies;
assigning a first set of edge weights to each task of the plurality of tasks based on the identified task dependencies, wherein at least one edge weight of the first set of edge weights is non-zero;
ordering the tasks based on the first set of edge weights to obtain a task order;
selecting a highest order task based on the task order, wherein the highest ordered task does not have a zero-edge weight;
executing the highest ordered task;
updating, dynamically, based on an execution of the highest order task, the first set of edge weights for remaining tasks to obtain an updated task order;
selecting a highest order task from the updated task order;
executing the highest ordered task from the updated task order, wherein tasks with a zero-edge weight are not executed; and
updating, dynamically, based on an execution of the highest order task from the updated task order, a second set of edge weights for remaining tasks in updated task order to obtain a second updated task order,
wherein at least one of the first set of edge weights is modified after the execution of the highest ordered task, and
wherein the task dependencies are generated by a compiler during compile-time.
10. The computing system of claim 9 ,
wherein the task dependencies are specified using a directed graph comprising a plurality of edges; and
wherein each of a first set of edges in the directed graph is associated with one of the first set of edge weights.
11. The computing system of claim 9 , wherein each of the task dependencies are specified using a directed graph.