Methods for avoiding overcommitment of virtual capacity in a redundant hierarchic data storage system

摘要

A hierarchic disk array data storage system (10) has a disk array (11) with multiple storage disks (12) and a disk array controller (14) which coordinates data transfer to and from the disks. The storage disks define a physical storage space (34). A RAID management system (16) maps a RAID-level virtual storage space (40) onto the physical storage space. The RAID-level virtual storage space presents the physical storage space as multiple RAID areas. The RAID areas include mirror RAID areas which contain mirror allocation blocks to store data according to mirror redundancy and parity RAID areas which contain parity allocation blocks to store data according to parity redundancy. The RAID management system (16) dynamically migrates data between the mirror and parity RAID areas in a manner which optimizes performance and data reliability. As new storage requests are made, the RAID management system evaluates the existing storage conditions and computes a total virtual capacity that is available to the user given a disk array configuration. The total virtual capacity is computed based upon a pessimistic assumption regarding a worst case data storage mix between mirror and parity RAID areas. The RAID management system will approve the new storage request without resorting to data migration if the new space requirements and already existing committed capacity does not exceed the total virtual capacity. In this manner, the RAID management system avoids over-commitment to the user.