Browse Prior Art Database

Method for storage-less scheduler to enforce "at least once within window" run policy

IP.com Disclosure Number: IPCOM000254561D
Publication Date: 2018-Jul-11
Document File: 1 page(s) / 10K

Publishing Venue

The IP.com Prior Art Database

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 69% of the total text.

Disclosed is a system for scheduling process that must be run with "at least once"

within specified time window policy. There's no control over execution environment (it

may be not available during part of the window, it can be turned off and then on at any

time). The process should never continue its execution beyond specified time window.

Process manager should keep execution state in memory. If execution

environment is available during the whole-time window, it should run the process at the

beginning of it. In case of execution environment being non-available (information about

previous process execution is lost), process should be run immediately after it becomes

available again, but only if it's still within time window. If process is running, but time

window is about to close, then process manager should stop the process execution.

Described solution has the following benefits: it enforces "at least once" policy

(when job execution time fits the time window minus scheduler downtime), there's no

need for additional job persistent storage and it does not allow job to run indefinitely

(contrary to most schedulers).

As an example, there is a cloud environment with a scheduler module deployed

onto it (as part of larger appliance). Scheduler runs computation of user usage statistics

daily. Time window was selected to be between midnight and 2 am, so there would be

least impact on normal system operations (process should not run after that time,

otherwise users may get longer...