Browse Prior Art Database

METHOD AND MEANS FOR ENHANCED DATA SCRUBBING OF RAID 0, AND 5 ARRAYS

IP.com Disclosure Number: IPCOM000013499D
Original Publication Date: 2000-Mar-01
Included in the Prior Art Database: 2003-Jun-18
Document File: 2 page(s) / 41K

Publishing Venue

IBM

Abstract

Disclosed is an algorithm and implementation in a Redundant Array Inexpensive Disk (RAID) controller's firmware to detect the growth of medium defects and to reconstruct the affected area in the array efficiently. The process of detecting and reconstructing is referred to as Data Scrubbing in this disclosure.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 1 of 2

  METHOD AND MEANS FOR ENHANCED DATA SCRUBBING OF RAID 0, AND 5 ARRAYS

Disclosed is an algorithm and implementation in a Redundant Array Inexpensive Disk (RAID) controller's firmware to detect the growth of medium defects and to reconstruct the affected area in the array efficiently. The process of detecting and reconstructing is referred to as Data Scrubbing in this disclosure.

In a RAID technology, multiple physical drives are bundled together to form what is referred to as an array. An array is configured into various RAID levels to enhance either performance or fault tolerance. The data are striped across all the physical drives which make up the logical drive. Therefore, when a medium defect grows on one of the physical drives, data availability is affected for the stripe unit spanning the medium defect for RAID
0. However, data availability is maintained for RAID 1, which mirrors its disk, and RAID 5, which reconstructs the data from its parity.

The Data Scrubbing detects a grown defect and fixes it in the background automatically and efficiently. In addition, the entire logical drive must be "scrubbed" within a reasonable time period at maximum system load. Data scrubbing time insures defect detection and correction when system operates at its peak for long periods of time. Having a minimum scrubbing time prevents over taxing the drives causing premature wear.

The efficiency of the algorithm stems from the use of the Small Computer System Interconnect (SCSI) Verify command for a large block of data. Such commands cause all the physical drives of the logical drive to read and verify a large block of data internally without transferring data across the SCSI bus. This eliminates taxing the RAID controller's Central Processing Unit (CPU) and peripheral hardware resources. In addition, all the logical drives are verified in parallel. If any of the physical drives encounter any grown defects, the physical drive will return the status indicating such defect. The defect repair routine, which uses more controller's resources, is then invoked to fix the affected spot on the drive.

For RAID 0, a SCSI Read command is issued to read the block of the data of the affected area from the drive. The effect of this read action could cause the drive to heal the defect due to the drive's capability of read reassignment. For RAID 1, the affected area of data and its mirrored data are read. The block of data from the good drive is written into the defected block of data. If bo...