Method and apparatus for a pattern based spaced management system

Claim 2

A pattern based space management apparatus for a computer system having at least two levels of hierarchical storage, said apparatus comprising: means for creating and maintaining a data base containing size information, name information, and historical information about the use of data sets residing on the computer system; first means for calculating a first next reference date for certain of the data sets and a first confidence level for each of said first next reference dates from information in said data base; second means for calculating a second next reference date for said certain of the data sets residing on the computer system and a second confidence level for each of said second next reference dates from the information in said data base including said name information; means for defining an amount of the highest level storage space which is to remain available; means for identifying, said means for identifying comparing said first and second confidence levels and selecting one of said first and second reference dates to be used based on which has the higher confidence level, said means identifying which data sets should be migrated between the storage levels of the computer system based on said selected next reference dates, the sizes of the data sets, and the amount of highest level storage space which is to remain available; and means for migrating the data sets identified by said means for identifying.

Claim 3

The apparatus of claim 1 wherein said data base includes age information about the data sets residing on the computer system, and wherein said second means for calculating calculates said second next reference date for said certain of the data sets residing on the computer system and said second confidence level for each of said second next reference dates from the information in said data base including said age information when there is no name information.

Claim 4

The apparatus of claim 1 wherein said means for identifying identifies files to be migrated from higher to lower levels of storage and identifies files to be migrated from lower to higher levels of storage.

Claim 5

The apparatus of claim 1 wherein said data base includes an access cycle table for storing access cycles for each data set, each access cycle including a number of active days and a number of idle days, said access cycle table also including a total cycle count indicating the total number of access cycles observed for each data set and a weighted count indicative of the number of times a particular access cycle is observed, said access cycle table being used by said first calculating means to calculate said first next reference date and said first confidence level.

Claim 6

The apparatus of claim 4 wherein said means for creating and maintaining includes means for comparing a current access cycle to said stored access cycles for a particular data set, means for incrementing said weighted count for that access cycle by the total cycle count for that data set in the event of a match, and means for saving the current access cycle and assigning it a weighted count equal to the total cycle count for that data set if a match is not found.

Claim 7

The apparatus of claim 5 wherein said means for saving stores the current access cycle and assigned weighted count in said access cycle table if said access cycle table has space available for that data set, and replaces the access cycle having the lowest weighted count if said access table has no more space available for that data set.

Claim 8

The apparatus of claim 4 wherein said first means for calculating includes means for adding the weighted counts from all the cycle table entries for a particular data set to determine a total weighted count, first means for scanning the cycle table entries from the highest weighted count to the lowest weighted count to locate the row in the access cycle table at which the scanned sum of weighted counts is at least ninety percent of the total weighted count, and second means for scanning upward from said located row to identify the row in the access cycle table in which the number of idle days meets or exceeds the number of idle days in the current access cycle.

Claim 9

The apparatus of claim 7 wherein said first means for calculating calculates said first next reference date by adding one plus the number of idle days from the identified row to the last reference date for the data set.

Claim 10

The apparatus of claim 8 wherein said first means for calculating calculates said first confidence level based on the output from said second means for scanning and said total cycle count.

Claim 11

The apparatus of claim 1 wherein said data base includes a name table for storing a count of the number of data sets found having a common prefix and the frequency of access to those data sets, said name table being used by said second means for calculating to calculate said second next reference dates and said second confidence levels.

Claim 12

The apparatus of claim 1 wherein said data base includes an age table for storing the number of data sets found, the age of each data set, and whether the data set was accessed since the last cycle, said age table being used by said second means for calculating to calculate said second next reference dates and said second confidence levels.

Claim 13

A pattern based space management method for use on a computer system having at least two levels of hierarchical storage, said method comprising the steps of: creating and maintaining a data base containing size information and an access cycle table for storing access cycles for each data set residing on the computer system, each access cycle including a number of active days and a number of idle days, said access cycle table also including a total cycle count indicating the total number of access cycles observed for each data set and a weighted count indicative of the number of times a particular access cycle is observed, and wherein the step of creating and maintaining a data base further includes the steps of: comparing a current access cycle to said stored access cycles for a particular data set; incrementing said weighted count for that access cycle by the total cycle count for that data set in the event of a match; and saving the current access cycle and assigning it a weighted count equal to the total cycle count for that data set if a match is not found; calculating a first next reference date for certain of the data sets and a first confidence level for each of said first next reference dates from information in said data base; defining an amount of the highest level storage space which is to remain available; identifying which data sets should be migrated between the storage levels of the computer system based on said first next reference dates, said first confidence levels, the sizes of the data sets, and the amount of highest level storage space which is to remain available; and migrating the identified data sets.

Claim 14

A pattern based space management method for use on a computer system having at least two levels of hierarchical storage, said method comprising the steps of: creating and maintaining a data base containing size information and an access cycle table for storing access cycles for each data set residing on the computer system, each access cycle including a number of active days and a number of idle days, said access cycle table also including a total cycle count indicating the total number of access cycles observed for each data set and a weighted count indicative of the number of times a particular access cycle is observed, and wherein the step of creating and maintaining a data base further includes the steps of: comparing a current access cycle to said stored access cycles for a particular data set; incrementing said weighted count for that access cycle by the total cycle count for that data set in the event of a match; and saving the current access cycle and assigning it a weighted count equal to the total cycle count for that data set if a match is not found; calculating a first next reference date for certain of the data sets and a first confidence level for each of said first next reference dates from information in said data base; defining an amount of the highest level storage space which is to remain available; identifying which data sets should be migrated between the storage levels of the computer system based on said first next reference dates, said first confidence levels, the sizes of the data sets, and the amount of highest level storage space which is to remain available; and migrating the identified data sets.

Claim 15

The method of claim 12 wherein said step of saving includes the steps of storing the current access cycle and assigned weighted count in said access cycle table if said access cycle table has space available for that data set, and replacing the access cycle having the lowest weighted count if said access table has no more space available for that data set.

Claim 16

The method of claim 12 wherein said step of calculating a first next reference includes the steps of: adding the weighted count values from all the cycle table entries for a particular data set to determine a total weighted count; scanning the cycle table entries from the highest weighted count to the lowest weighted count to locate the row in the access cycle table at which the scanned sum of weighted counts is at least ninety percent of the total weighted count; scanning upward from said located row to identify the row in the access cycle table in which the number of idle days meets or exceeds the number of idle days of the current access cycle; and adding one plus the number of idle days from the identified row to the last reference date for the data set.

Claim 17

The method of claim 14 wherein the step of calculating said first confidence level includes the step of setting the confidence level to zero if the total cycle count for the data set is less than a first predetermined number.

Claim 18

The method of claim 14 wherein the step of calculating said first confidence level includes the step of setting the confidence level to zero if the data set has a number of idle days in the current cycle that is greater than the number of idle days recorded in the access cycle table for that data set.

Claim 19

The method of claim 14 wherein the step of calculating said first confidence level includes the step of setting the confidence level to zero if the data set is still in the active portion of its cycle.

Claim 20

The method of claim 14 wherein the step of calculating said first confidence level includes the step of setting the confidence level to 100% if the upward scanning step produces a match.

Claim 21

The method of claim 18 wherein the step of calculating said first confidence level includes the step of decreasing the confidence level from 100% by 2% for every cycle less than ten.