System for performing action by sorting actions into immediate and deferred queues, processing immediate queue while still sorting, and appending deferred queue to immediate after sorting

• US 5,623,693 A
• Filed: 02/17/1994
• Issued: 04/22/1997
• Est. Priority Date: 02/17/1994
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Ashton, Lyn L., Pearson, Anthony S., Pence, Jerry W.
• Primary Examiner(s): Lee, Thomas C.
• Assistant Examiner(s): Chen, Anderson I.
• Application Number:
  US08/198,005
• Time in Patent Office: 1,160 Days
• Field of Search: 395/275, 395/200, 395/650, 395/600, 395/859, 395/874, 364/148, 364/402, 370/94.1, 361/238.3, 361/280
• US Class Current: 710/5
• CPC Class Codes: G06F 2003/0697 device management, e.g. handlers, drivers, I/O schedulers
• CPC Class Codes: G06F 3/0601 Interfaces specially adapted for storage systems

Abstract

A procedure for the optimal processing of variable-cost actions such as encountered in the storage reclamation procedures for a multivolume data library. The procedure introduces a temporary processing queue to minimize idle processing capacity during the scanning and sorting of a large plurality of variable-cost actions such as the recycling of a plurality of data storage volumes each having a variable recycle processing cost related to the action of valid data remaining on the volume. Volumes (actions) are selected for the immediate queue according to a dynamically-adjusted threshold test for the processing cost. This processing cost threshold is dynamically adjusted to optimize the immediate queue in relation to the available processing capacity. After scanning and sorting all volumes according to recycle processing cost, the temporary (immediate) queue is updated to a final recycle processing queue by appending a sorted deferred queue to the remainder of the immediate queue. The procedure of this invention minimizes idle processing capacity during the queue-building interval, thereby optimizing the number of recovered data storage volumes released in a given time interval.

Claims

1. In a procedural environment having an action processing capacity and a plurality of N actions to perform, wherein the actions may be performed in any order, wherein a variable cost F_n =(0,1) can be assigned for each said action, wherein F_n =0 represents the lowest cost, and wherein N and n≦

N are positive integers, a machine-implemented method for processing said actions, said method comprising the steps of;

(a) arranging a plurality of M≦

N of said actions into a processing queue ordered by said variable-cost F_m, wherein M and m≦

M are positive integers, said arranging step comprising the steps of;

(a.1) apportioning said M actions into an immediate queue and a deferred queue by performing steps comprising;

(a.1.1) defining the variable-cost threshold T=[0,1];

(a.1.2) placing the m^th said action in said immediate queue if F_m ≦

T;

(a.1.3) placing the m^th said action at the beginning of said immediate queue if F_m =0; and

(a.1.4) otherwise placing said m^th action in said deferred queue;

(a.2) sorting said actions in order of said variable cost F_m in said deferred queue after completion of said apportioning step (a.1); and

(a.3) appending said deferred queue to said immediate queue to represent said processing queue after completion of said sorting step (a.2);

(b) before completion of said apportioning step (a.1) performing repeatedly the steps of

(b.1) performing to completion the first action in said immediate queue, and

(b.2) removing said action from said processing queue.

2. The method of claim 1 wherein said apportioning of step (a.1) further comprises the step of:

(a.1.5) adjusting said threshold T according to the relationship of the contents of said immediate queue to said action processing capacity.

3. The method of claim 2 wherein said adjusting step (a.1.5) comprises the steps of:

(a.1.5.1) adding an increment to increase said threshold T responsive to an inadequate number of said actions in said immediate queue;

(a.1.5.2) subtracting a decrement to decrease said threshold T responsive to the presence of an adequate plurality of said actions in said immediate queue; and

(a.1.5.3) multiplying by a fraction to decrease said threshold T responsive to the presence of an excessive plurality of said actions in said immediate queue.

4. In a data storage library having a recycle processing capacity and a plurality N of volumes for storing data, wherein a fraction F_n =[0,1] of the n^th said volume contains valid data and wherein N and n≦

N are positive integers, a machine-implemented method for compacting said valid data to recover empty volumes, said method comprising the steps of;

(a.1) arranging a plurality M≦

N of said volumes into a recycle processing queue ordered by said fraction F_m, wherein M and m≦

M are positive integers, said arranging step comprising the steps of;

(a.1.1) apportioning said M volumes into an immediate queue and a deferred queue by performing steps comprising;

(a.1.2) defining a valid data occupancy fraction threshold T=[0,1];

(a.1.3) placing the m^th said volume in said immediate queue if F_m ≦

T;

queue if F_m =0; and

(a.1.4)otherwise placing said m^th volume in said deferred queue;

(a.2) sorting said volumes in order of said faction F_m in said deferred queue after completion of said apportioning step (a.1); and

(a.3) appending said deferred queue to said immediate queue to form said recycle processing queue after completion of said sorting step (a.2);

(b) before completion of said apportioning step (a.1), performing repeatedly the steps of

(b.1) transferring said valid data from the first volume in said immediate queue to a compaction target volume, and

(b.2) removing said first volume from said recycle processing queue.

5. The method of claim 4 wherein said apportioning step (a.1) further comprises the step of:

(a.1.5) adjusting said threshold T according to the relationship of the contents of said immediate queue to said recycle processing capacity.

6. The method of claim 5 wherein said adjusting step (a.1.5) comprises the steps of:

(a.1.5.1) adding an increment to increase said threshold T responsive to an inadequate number of volumes in said immediate queue;

(a.1.5.2) multiplying by a fraction to decrease said threshold T responsive to the presence of an excessive plurality of volumes in said immediate queue.

7. In an apparatus for automatic management of a data storage library having a recycle processing capacity and a plurality N of volumes for storing data, wherein a fraction F_n =[0,1] of the n^th said volume contains valid data and wherein N and n≦

N are positive integers, a storage recovery system for compacting said valid data to recover empty volumes for reuse, said system comprising;

queuing means for arranging a plurality M≦

N of said volumes into a recycle processing queue ordered by said fraction F_m, wherein M and m≦

M are positive integers;

recycling means coupled to said queuing means for transferring said valid data from the first volume in said recycle processing queue to a compaction target volume and for removing said first volume from said recycle processing queue;

selection means in said queuing means for apportioning said plurality M of volumes into an immediate queue and a deferred queue;

launching means coupled to said queuing means and said recycling means for starting said recycling means before formation of said recycle processing queue, whereby said first volume is taken from said immediate queue;

sorting means in said queuing means for sorting the volumes in order of said fraction F_m =[0,1] in said deferred queue and for appending said deferred queue to said immediate queue to firm said recycle processing queue;

thresholding means coupled to said selection means for comparing said fraction F_m for the m^th said volume with a valid data occupancy fraction threshold T=[0,1], whereby said m^th volume is added to said immediate queue if F_m ≦

T and otherwise is added to said deferred queue; and

bypass means coupled to said thresholding means for adding said m^th volume to the front of said immediate queue when F_m =0.

8. The storage recovery system of claim 7 further comprising:

adjusting means coupled to said thresholding means for adjusting said threshold T according to the relationship of the contents of said immediate queue to said recycle processing capacity.

9. The storage recovery system of claim 8 further comprising:

adding means in said adjusting means for adding an increment to increase said threshold T responsive to the presence of an inadequate plurality of volumes in said immediate queue; and

proportioning means in said adjusting means for multiplying by a fraction to decrease said threshold T responsive to the presence of an excessive plurality of volumes in said immediate queue.

10. An automated data storage tape library having a recycle processing capacity and a plurality N of tape volumes for storing data, wherein a fraction F_n =[0,1] of the n^th said tape volume contains valid data and wherein N and n≦

N are positive integers, said tape library including volume recovery means for compacting said valid data to recover empty tape volumes for reuse, wherein said volume recovery means comprises;

queuing means for arranging a plurality M≦

N of said tape volumes into a recycle processing queue ordered by said fraction F_m, wherein M and m≦

M are positive integers;

recycling means coupled to said queuing means for transferring said valid data from the first tape volume in said recycle processing queue to a compaction target volume and for removing said first tape volume from said recycle processing queue;

selection means in said queuing means for apportioning said plurality M of tape volumes into an immediate queue and a deferred queue;

launching means coupled to said queuing means and said recycling means for starting said recycling means before formation of said recycle processing queue, whereby said first tape volume is taken from said immediate queue;

sorting means in said queuing means for sorting the tape volumes in order of said fraction F_m =[0,1] in said deferred queue and for appending said deferred queue to said immediate queue to form said recycle processing queue;

thresholding means coupled to said selection means for comparing said fraction F_m for the m^th said tape volume with a valid data fraction occupancy threshold T=[0,1], whereby said m^th tape volume is added to said immediate queue if F_m ≦

T and otherwise is added to said deferred queue; and

bypass means coupled to said thresholding means for adding said m^th tape volume to the front of said immediate queue when F_m =0.

11. The automated data storage tape library of claim 10 further comprising:

adjusting means coupled to said thresholding means for adjusting said threshold T according to the relationship of the contents of said immediate queue to said recycle processing capacity.

12. The automated data storage tape library of claim 11 further comprising:

adding means in said adjusting means for adding an increment to increase said threshold T responsive to the presence of an inadequate plurality of tape volumes in said immediate queue; and

proportioning means in said adjusting means for multiplying by a fraction to decrease said threshold T responsive to the presence of an excessive plurality of tape volumes in said immediate queue.

13. An automated optical data storage library having a recycle processing capacity and a plurality N of optical volumes for storing data, wherein a fraction F_n =[0,1] of the n^th said optical volume contains valid data and wherein N and n≦

N are positive integers said optical data storage library including volume recovery means for compacting said valid data to recover empty optical volumes for reuse wherein said volume recovery means comprises;

queuing means for arranging a plurality M≦

N of said optical volumes into a recycle processing queue ordered by said fraction F_m, wherein M and m≦

M are positive integers;

recycling means coupled to said queuing means for transferring said valid data tom the first optical volume in said recycle processing queue to a compaction target volume and for removing said first optical volume from said recycle processing queue;

selection means in said queuing means for apportioning said plurality M of optical volumes into an immediate queue and a deferred queue;

launching means coupled to said queuing means and said recycling means for starting said recycling means before formation of said recycle processing queue, whereby said first optical volume is taken from said immediate queue;

sorting means in said queuing means for sorting the optical volumes in order of said fraction F_m =[0,1] in said deferred queue and for appending said deferred queue to said immediate queue to form said recycle processing queue;

thresholding means coupled to said selection means for comparing said fraction F_m for the m^th said optical volume with a valid data fraction occupancy threshold T=[0,1], whereby said m^th optical volume is added to said immediate queue if F_m ≦

T and otherwise is added to said deferred queue; and

bypass means coupled to said thresholding means for adding said m^th optical volume to the front of said immediate queue when F_m =0.

14. The automated optical data storage library of claim 13 further comprising:

adjusting means coupled to said thresholding means for adjusting said threshold T according to the relationship of the contents of said immediate queue to said recycle processing capacity.

15. The automated optical data storage library of claim 14 further comprising:

adding means in said adjusting means for adding an increment to increase said threshold T responsive to the presence of an inadequate plurality of optical volumes in said immediate queue; and

proportioning means in said adjusting means for multiplying by a fraction to decrease said threshold T responsive to the presence of an excessive plurality of optical volumes in said immediate queue.

16. A machine-implemented method for reclaiming data storage volumes in a multi-volume data storage library having a plurality of said data storage volumes, each volume having a variable-cost of reclaiming valid data thereon, said method comprising the steps of:

arranging a plurality of data storage volumes into a processing queue and ordering said volumes according to their variable-costs by performing steps comprising;

apportioning the volumes into an immediate queue and a deferred queue by performing steps comprising;

defining a variable-cost threshold;

for each data storage volume, performing steps comprising;

placing the volume in the immediate queue if its variable-cost does not exceed the variable-cost threshold;

otherwise placing the volume in the deferred queue; and

reconfiguring the processing queue by sorting the volumes in the deferred queue in order of variable cost and then appending the sorted deferred queue to the immediate queue; and

repeatedly performing the steps of (a) copying valid data from the first volume in the processing queue to a designated compaction target volume, and (b) removing the first volume from said processing queue.

17. The method of claim 16, wherein the repeatedly performed steps are initiated before completion of the apportioning step.

18. The method of claim 16, further comprising steps of adjusting the variable-cost threshold according to the relationship of the contents of said immediate queue to an action processing capacity of said storage library, said adjusting steps comprising:

responsive to an inadequate number of volumes in the immediate queue, increasing the variable-cost threshold by a predetermined increment;

responsive to the presence of an adequate plurality of volumes in the immediate queue, decreasing the variable-cost threshold by a predetermined increment; and

responsive to the presence of an excessive plurality of volumes in the immediate queue, decreasing the variable-cost threshold by multiplying the variable-cost threshold by a fraction less than one.

19. The method of claim 16, the variable cost of each data volume being a representation of the data occupancy of that volume.

20. The method of claim 16, the apportioning step further comprising the step of, for each data storage volume, placing the volume at the beginning of the immediate queue if its variable-cost is substantially zero.