Recovery from errors in a redundant array of disk drives

• US 5,278,838 A
• Filed: 06/18/1991
• Issued: 01/11/1994
• Est. Priority Date: 06/18/1991
• Status: Expired

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Palmer, David W., Thompson, Richard S., Ng, Spencer W.
• Primary Examiner(s): Canney, Vincent P.
• Application Number:
  US07/717,263
• Time in Patent Office: 938 Days
• Field of Search: 371/10.1, 371/11.1, 371/40.2, 371/12, 371/66, 345/578, 364/268.5, 364/269.2
• US Class Current: 714/6.12
• CPC Class Codes: G06F 11/1092 Rebuilding, e.g. when physically replacing a failing disk
• CPC Class Codes: G11B 20/1833 by adding special lists or symbols to the coded information G11B20/1806, G11B20/1866 take precedence

Abstract

Fault tolerance in a redundant array of disk drives is degraded when error conditions exist in the array. Several methods for rebuilding data of the array to remove the degradation are described. Data rebuilding for entire disk drives and partial data rebuilds of disk drives are described. All rebuild methods tend to reduce the negative affect of using array resources for the data rebuild. In one method rebuilding occurs during idle time of the array. In a second method rebuilding is interleaved between current data area accessing operations of the array at a rate which is inversely proportional to activity level of the array. In a third method, the data are rebuilt when a data area being accessed is a data area needing rebuilding.

Claims

1. In a machine-effected method of rebuilding data in a redundant array of a plurality of disk drives which includes an error affected disk drive, including the machine-executed steps of:

detecting and indicating that a one of the disk drives is error affected;

measuring and indicating a rate of machine operations of the array;

establishing a rate of rebuilding data affected by the error affected disk drive which rate is predetermined inversely proportional to said measured and indicated rate of accesses;

intermediate predetermined ones of said accesses which is in said inverse proportion, rebuilding data in a predetermined one of the disk drives for replacing data in error.

2. In the machine-effected method set forth in claim 1 further including the machine-executed steps of:

accessing the error affected disk drive;

during said access rebuilding data affected by said error.

3. In the machine-effected method set forth in claim 1 further including the machine-executed steps of:

detecting that the array of disk drives has no current access; and

rebuilding data affected by said error affected disk drive beginning upon said detection of no current access.

4. In a machine-effected method of automatically maintaining fault tolerance in a parity array of disk drives including the machine-executed steps of:

detecting and indicating a degradation of the fault tolerance of the parity array;

evaluating and indicating the current information handling activity of the parity array;

establishing a plurality of data rebuild methods for the parity array for removing the fault tolerance degradation from the parity array; and

analyzing the indicated current information handling activity of the parity array and selecting a one of the plurality of rebuild methods which effects a data rebuild without degrading performance of said current information handling activity more than a predetermined degradation level.

5. In the machine-effected method set forth in claim 4 further including the machine-executed steps of:

in said evaluating step, determining the rate of information handling activity and selecting a data rebuild rate in a predetermined inverse ratio to the determined rate of information handling activity; and

establishing a one of the plurality of rebuild methods as a variable rate rebuild method which effects data rebuilding at said selected rebuild rate of a predetermined number of addressable error-affected data units in the parity array.

6. In the machine-effected method set forth in claim 5 further including the machine-executed steps of:

completing a data rebuild using said variable rebuild method;

detecting that the parity array is idle; and

continuing the data rebuilding of additional ones of the addressable error-affected data units so long as the parity array is idle.

7. In the machine-effected method set forth in claim 5 further including the machine-executed steps of:

in said evaluating step, determining that the parity array is idle; and

establishing a second one of the plurality of rebuild methods as an idle rebuild method to be selected whenever the parity array is idle and a rebuild need exists in the parity array.

8. In the machine-effected method set forth in claim 5 further including the machine-executed steps of:

performing a data area access operation in the parity array;

while performing the data area access operation, performing said detecting and indicating step for detecting and indicating that the data access operation is accessing a one of the addressable data units needing a data rebuild; and

rebuilding at least the data unit being accessed.

9. In a machine-effected method of automatically maintaining fault tolerance in a fault tolerant parity array of disk drives including the machine-executed steps of:

detecting that the parity array is idle;

detecting that the parity array fault tolerance is degraded and needs data rebuilding;

identifying data rebuild needs; and

rebuilding the data at the identified rebuild needs during said detected idle times.

10. In a machine-effected method of automatically maintaining fault tolerance in a fault tolerant parity array of disk drives including the machine-executed steps of:

indicating that fault tolerance of the parity array is degraded by a plurality of error-affected addressable data units of the parity array which respectively need data rebuilding to reestablish the fault tolerance;

performing a data area access operation to an addressable data unit in the parity array;

while performing the data area access operation, detecting and indicating that the data access operation is accessing a one of the error-affected addressable data units needing a data rebuild; and

rebuilding the addressable data unit being accessed.

11. Apparatus having a redundant array of disk devices, the improvement including, in combination:

rebuild need evaluation means for detecting and indicating a degradation in the redundant array including indicating a one of the disk drives needs to have data rebuilt to such one disk drive;

access rate means for measuring and indicating a rate of machine operations of said array;

rebuild rate means coupled to said evaluation means and to said access rate means for responding to said indicated rebuild need and to said indicated operations rate for establishing and indicating a predetermined rate of rebuilding for the array for recovering from said degradation of fault tolerance; and

rebuild means having a plurality of data rebuild effecting means and being coupled to said rebuild rate means and to said rebuild need means for effecting data rebuild in said one disk drive using a predetermined one of said plurality of data rebuild effecting means.

12. Apparatus having a redundant array of disk drives as set forth in claim 11, further including, in combination:

control means in the apparatus for controlling access to the disk drives in the redundant array and for detecting and indicating when the array is currently not being accessed for a data handling operation;

a first one of said rebuild effecting means being connected to said control means, to said rebuild rate means and to said rebuild need means for determining a rebuild can be scheduled and then activating the control means to give access to the redundant array to the first one of said rebuild effecting means for effecting a series of time space-apart rebuild operations at said predetermined rate.

13. Apparatus having a redundant array of disk drives as set forth in claim 12, further including, in combination:

said rebuild rate means including means or indicating a plurality of rebuild rates, said plurality of rebuild rates increasing in rate values in an inverse proportion to said indicated operations rate and said rate values corresponding respectively to predetermined ranges of said indicated machine operations rates; and

predetermined rate means in said rebuild rate means for indicated said predetermined rate as one of said plurality of rebuild rates which corresponds to a current one of the indicated machine operations rate.

14. Apparatus having a redundant array of disk drives as set forth in claim 11, further including, in combination:

control means in the apparatus for controlling access to the disk drives in the redundant array and for detecting and indicating when the array is currently not being accessed for a data handling operation;

a second one of said rebuild effecting means being connected to said control means indicating the array is not being currently accessed to actuate the control means to give said second one rebuild effecting means access to the redundant array for starting a predetermined data rebuild in response to said not being accessed indication.

15. Apparatus having a redundant array of disk drives as set forth in claim 11, further including, in combination:

control means in the apparatus for controlling access to the disk drives in the redundant array and for detecting and indicating when the array is currently not being accessed for a data handling operation;

a third one of said plurality of rebuild effecting means being connected to rebuild need evaluation means and to said control means for receiving indication that a disk drive access is occurring in a given area of the array which needs a data rebuild indicated by said rebuild need evaluation means and responding to said received indication of said access to effect a data rebuild of a predetermined area of the redundant array which includes said given area.

16. Apparatus having a redundant array of disk drives as set forth in claim 11, further including, in combination:

control means in the apparatus for controlling access to the disk drives in the redundant array and for detecting and indicating when the array is currently not being accessed for a data handling operation and whether or not any pending access requests are currently pending;

a first one of said rebuild effecting means being connected to said control means, to said rebuild rate means and to said rebuild need means for determining a rebuild can be scheduled and then activating the control means to give access to the redundant array to the first one of said rebuild effecting means for effecting a series of time space-apart rebuild operations at said predetermined rate;

a second one of said rebuild effecting means being connected to said control means and being responsive to the control means indicating the array is not being currently accessed and there are no access requests pending to actuate the control means to give said second one rebuild effecting means access to the redundant array for starting a predetermined data rebuild in response to said not being accessed indication;

a third one of said plurality of rebuild effecting means being connected to rebuild need evaluation means and to said control means for receiving indication that a disk drive access is occurring in a given area of the array which needs a data rebuild indicated by said rebuild need evaluation means and responding to said received indication of said access to effect a data rebuild of a predetermined area of the redundant array which includes said given area; and

each of said effecting means upon completing a predetermined data rebuild operation being responsive to the control means then indicating no current access or no pending access requests to initiate another predetermined data rebuild operation.