Server resource balancing using a dynamic-sharing strategy

A system comprising:

• one or more computers;
• and
• a non-transitory computer-readable medium coupled to the one or more computers having instructions stored thereon which, when executed by the one or more computers, cause the one or more computers to perform operations comprising:
• receiving, in a dispatching layer, a request to run a first task for a first application, the request including a first application priority;
• identifying at least one second application priority of at least one currently running application, wherein the at least one second application priority is different than the first application priority;
• determining a maximum number of allowable parallel tasks per application priority;
• assigning application priority weights to each of the first application priority and the at least one second application priority, wherein a first application priority weight is assigned to the first application priority for the first application and at least one second application priority weight is assigned to the at least one second application priority for the at least one currently running application;
• after assigning the application priority weights, determining a first application priority divisor for the first application priority and a second application priority divisor for the at least one second application priority, wherein the first application priority divisor is determined by multiplying the first application priority weight by a first number of currently running applications of the first application priority, and the second application priority divisor for the at least one second application priority is determined by multiplying the at least one second application priority weight by a second number of currently running applications of the at least one second application priority;
• after determining the first application priority divisor and the second application priority divisor, determining an overall divisor based on the first application priority divisor and the second application priority divisor;
• after determining the overall divisor, determining a number of parallel tasks for the first application and the at least one currently running application based on the maximum number of allowable parallel tasks per application priority, the overall divisor, and a respective application priority weight, wherein determining includes assigning a first number of parallel tasks to the first application;
• and
• executing the first application using the assigned first number of parallel tasks.

A computer program product encoded on a non-transitory storage medium, the product comprising non-transitory, computer readable instructions for causing one or more processors to perform operations comprising:

• receiving, in a dispatching layer, a request to run a first task for a first application, the request including a first application priority;
• identifying at least one second application priority of at least one currently running application, wherein the at least one second application priority is different than the first application priority;
• determining a maximum number of allowable parallel tasks per application priority;
• assigning application priority weights to each of the first application priority and the at least one second application priority, wherein a first application priority weight is assigned to the first application priority for the first application and at least one second application priority weight is assigned to the at least one second application priority for the at least one currently running application;
• after assigning the application priority weights, determining a first application priority divisor for the first application priority and a second application priority divisor for the at least one second application priority, wherein the first application priority divisor is determined by multiplying the first application priority weight by a first number of currently running applications of the first application priority, and the second application priority divisor for the at least one second application priority is determined by multiplying the at least one second application priority weight by a second number of currently running applications of the at least one second application priority;
• after determining the first application priority divisor and the second application priority divisor, determining an overall divisor based on the first application priority divisor and the second application priority divisor;
• after determining the overall divisor, determining a number of parallel tasks for the first application and the at least one currently running application based on the maximum number of allowable parallel tasks per application priority, the overall divisor, and a respective application priority weight, wherein determining includes assigning a first number of parallel tasks to the first application;
• and
• executing the first application using the assigned first number of parallel tasks.