Reliability-aware application scheduling

One or more computer-readable storage media comprising computer-executable instructions for scheduling processing jobs on processing entities, the computer-executable instructions, when executed by one or more processing units of a computing device, cause the computing device to perform steps comprising:

• receiving real-time reliability data associated with a processing entity;
• identifying a statistical reliability model that correlates processing, entity failures to factors quantified by the received real-time reliability data;
• generating, with the identified statistical reliability model, a predicted future reliability of the processing entity based on at least some of the received real-time reliability data;
• generating a reliability score for the processing entity based on both the predicted future reliability of the processing entity and a prior reliability score for the processing entity;
• receiving a processing job for scheduling, the processing job having associated with it a minimum acceptable reliability score;
• and
• scheduling the received processing job on the processing entity only if the generated reliability score is greater than or equal to the minimum acceptable reliability score associated with the processing job.

A system for executing processing jobs, the system comprising:

• multiple collections of computing devices;
• a first reliability-aware scheduler computing device performing steps comprising:
• receiving real-time reliability data associated with a first collection of computing devices from among the multiple collections of computing devices;
• generating a reliability score for the first collection of computing devices based on at least some of the received reliability data associated with the first collection of computing devices and based on a prior reliability score for the first collection of computing devices;
• receiving a processing job for scheduling, the processing job having associated with it a minimum acceptable reliability score;
• and
• scheduling the received processing job on the first collection of computing devices only if the generated reliability score of the first collection of computing devices is greater than or equal to the minimum acceptable reliability score associated with the processing job;
• and
• a second reliability-aware scheduler computing device performing steps comprising:
• receiving real-time reliability data associated with a first computing device in the first collection of computing devices;
• generating a reliability score for the first computing device based on at least some of the received reliability data associated with the first computing device and based on a prior reliability score for the first computing device;
• receiving, from the first reliability-aware scheduler computing device, the processing job for scheduling on a computing device of the first collection of computing devices;
• and
• scheduling the received processing job on the first computing device only if the generated reliability score of the first computing device is greater than or equal to the minimum acceptable reliability score associated with the processing job.