Self-governed contention-aware approach to scheduling file defragmentation

摘要

A method, system, and computer program product for file storage defragmentation on a cluster of nodes. The method for self-governed, contention-aware scheduling of file defragmentation operations commences by calculating a score for candidate files of a storage volume, where the score is based on a fragmentation severity value. The process proceeds to determine an amount of contention for access to a candidate file (e.g., by accessing the candidate file to record the amount of time it takes to obtain access). If the fragmentation severity value and the amount of contention suggestion a benefit from defragmentation, then the method initiating defragmentation operations on the candidate file. The method delays for a calculated wait time before performing a second defragmentation operation. Real-time monitors are used to determine when the contention is too high or when system utilization is too high. Only files that have ever been opened are considered candidates for defragmentation.

主权项

1. A computer implemented method for self-governed, contention-aware scheduling of file defragmentation, comprising:
selecting a plurality of candidate files of a storage volume for performing defragmentation operations, using a score calculated by combining a frequency or recency of access and a fragmentation severity value of a respective candidate file; executing, by a computer, a process to determine an amount of contention of access to a candidate file of the plurality of candidate files, wherein the amount of contention indicates an amount of time to obtain access to the candidate file; initiating a defragmentation operation on the candidate file if the contention is determined to be below a contention threshold value; initiating the defragmentation operation after delaying, for a calculated wait time, the defragmentation operation on the candidate file if the amount of time to obtain access to the candidate file is above the contention threshold value; determining, after initiating the defragmentation operation on the candidate file, a system utilization value; and initiating a second defragmentation operation if the system utilization value is below a headroom threshold value, and delaying the initiation if the system utilization value is above the headroom threshold value.