Fault tolerant disk drive system with error detection and correction

Claim 1

A fault tolerant disk drive system comprising: a first plurality of disk drive devices for storing data; means for performing a first plurality of parity checks including a plurality of vertical parity checks, and a plurality of diagonal parity checks on data coupled to said first plurality of disk drive devices; a second plurality of disk drive devices for storing results of said first plurality of parity checks; means for performing a second plurality of parity checks including a plurality of vertical parity checks and a plurality of diagonal parity checks on said data in the event said data is outputted from said first plurality of disk drive devices; means for comparing results from said first plurality of parity checks and said second plurality of parity checks and generating comparison results; means for utilizing the comparison results and unaffected data to detect and correct data errors.

Claim 8

A method for detecting corrupted data in a disk drive array, said disk drive array comprising a plurality of data disk drives, said method comprising the steps of: coupling a plurality of data bits in parallel to said plurality of data disk drives; performing a first plurality of read-in parity checks on said plurality of data bits, said first plurality of read-in parity checks comprises a plurality of read-in vertical parity checks; performing a second plurality of read-in parity checks on said plurality of data bits, said second plurality of read-in parity checks comprises a plurality of read-in diagonal parity checks; storing said plurality of data bits in said plurality of data disk drives; outputting said plurality of data bits from said plurality of data disk drives; performing a first plurality of read-out parity checks on said plurality of data bits outputted from said plurality of disk drives, said first plurality of read-out parity checks comprises a plurality of read-out vertical parity checks; performing a second plurality of read-out parity checks on said plurality of data bits outputted from said plurality of disk drives, said plurality of read-out parity checks comprises a plurality of read-out diagonal parity checks; utilizing results of said first plurality of read-in parity checks, said second plurality of read-in parity checks, said first plurality of read-out parity checks, and said second plurality of read-out parity checks to detect corrupted data; and identification stamping each block of said data bits of each said data disk drives each time a write occurs, said identification being used to overcome reliability problems occurring in said disk drive array.

Claim 9

A method for detecting corrupted data in a disk drive array, said disk drive array comprising a plurality of data disk drives, said method comprising the steps of: coupling a plurality of data bits in parallel to said plurality of data disk drives; performing a first plurality of read-in parity checks on said plurality of data bits, said first plurality of read-in parity checks comprises a plurality of read-in vertical parity checks; performing a second plurality of read-in parity checks on said plurality of data bits, said second plurality of read-in parity checks comprises a plurality of read-in diagonal parity checks; storing said plurality of data bits in said plurality of data disk drives; outputting said plurality of data bits from said plurality of data disk drives; performing a first plurality of read-out parity checks on said plurality of data bits outputted from said plurality of disk drives, said first plurality of read-out parity checks comprises a plurality of read-out vertical parity checks; performing a second plurality of read-out parity checks on said plurality of data bits outputted from said plurality of disk drives, said plurality of read-out parity checks comprises a plurality of read-out diagonal parity checks; utilizing results of said first plurality of read-in parity checks, said second plurality of read-in parity checks, said first plurality of read-out parity checks, and said second plurality of read-out parity checks to detect corrupted data; and identification stamping each block of said data bits of each said data disk drives each time a write occurs, said identification being used to overcome reliability problems occurring in said disk drive array.

Claim 15

A method for locating a corrupted data in N data disk drives, in a system comprising data disk drives, data buffer means having a corresponding number N output channels for providing data to said N data disk drives, and first and second redundant disk drives among said N data disk drives, said method comprising the steps of: outputting a first column of N data bits from said data buffer means; performing a first read-in vertical parity check on said first column of N data bits; storing said first column of N data bits in said N data disk drives; storing result of said first read-in vertical parity check in said first redundant disk drive; outputting a second column of N data bits from said data buffer means; performing a read-in diagonal parity check on a bit in a first channel of said first column of N data bits; performing a second read-in vertical parity check on said second column of N data bits and result of said read-in diagonal parity check; storing said second column of N data bits in said N data disk drives; storing result of said second read-in vertical parity check in said first redundant disk drive; storing the result of said read-in diagonal parity check in said second redundant disk drive; reading out said first column of N data bits from said N data disk drives; reading out said result of said first read-in vertical parity check from said first redundant disk drive; performing a first read-out vertical parity check on said first column of N data bits read out from said N data disk drives; reading out said second column of N data bits from said N data disk drives; reading out said result of said second read-in vertical parity check from said first redundant disk drive; reading out said result of said read-in diagonal parity check from said second redundant disk drive; performing a second read-out vertical parity check on said second column of N data bits read out from said N data disk drives; performing a read-out vertical parity check on a bit in said first channel of said first column of N data bits read out from said N data disk drives; utilizing the results of said first read-in vertical parity check, said first read-out vertical parity check, said second read-in vertical parity check, said second read-out vertical parity check, said read-in vertical parity check, and said read-out vertical parity check to locate corrupted data in said N data disk drives.

Claim 16

A method for locating a corrupted data in N data disk drives, in a system comprising data disk drives, data buffer means having a corresponding number N output channels for providing data to said N data disk drives, and first and second redundant disk drives among said N data disk drives, said method comprising the steps of: outputting a first column of N data bits from said data buffer means; performing a first read-in vertical parity check on said first column of N data bits; storing said first column of N data bits in said N data disk drives; storing result of said first read-in vertical parity check in said first redundant disk drive; outputting a second column of N data bits from said data buffer means; performing a read-in diagonal parity check on a bit in a first channel of said first column of N data bits; performing a second read-in vertical parity check on said second column of N data bits and result of said read-in diagonal parity check; storing said second column of N data bits in said N data disk drives; storing result of said second read-in vertical parity check in said first redundant disk drive; storing the result of said read-in diagonal parity check in said second redundant disk drive; reading out said first column of N data bits from said N data disk drives; reading out said result of said first read-in vertical parity check from said first redundant disk drive; performing a first read-out vertical parity check on said first column of N data bits read out from said N data disk drives; reading out said second column of N data bits from said N data disk drives; reading out said result of said second read-in vertical parity check from said first redundant disk drive; reading out said result of said read-in diagonal parity check from said second redundant disk drive; performing a second read-out vertical parity check on said second column of N data bits read out from said N data disk drives; performing a read-out vertical parity check on a bit in said first channel of said first column of N data bits read out from said N data disk drives; utilizing the results of said first read-in vertical parity check, said first read-out vertical parity check, said second read-in vertical parity check, said second read-out vertical parity check, said read-in vertical parity check, and said read-out vertical parity check to locate corrupted data in said N data disk drives.