Two-dimensional storage array with prompt parity in one dimension and delayed parity in a second dimension

A data storage system, comprising:

• a plurality of disk drives logically arranged to form a two-dimensional disk drive array having a m+1 rows and n+1 columns where m≧2 and n≧2, and wherein each disk drive comprises q data storage regions where q≧1, and wherein each row of the disk drive array comprises q row stripes, and wherein each row stripe includes a different one of the q data storage regions of each of the n+1 disk drives in the same row, and wherein each column of the disk drive array comprises q column stripes, and wherein each column stripe includes a different one of the q data storage regions of each of the m+1 disk drives in the same column;
• a disk array controller coupled to each of the plurality of disk drives and configured to group data to be stored in a given row stripe during a write operation, to calculate a row parity block for the given row stripe, and to calculate a column parity block for each column stripe modified during the write operation;
• and
• wherein the disk array controller is configured to either:
• (i) calculate the row parity block for the given row stripe before completing the write operation, and delay the column parity block calculations until after the completion of the write operation, or (ii) calculate the column parity blocks before completing the write operation, and delay the row parity block calculation until after the completion of the write operation.

A method for storing data within a plurality of disk drives logically arranged to form a two-dimensional disk drive array having a plurality of rows and columns, wherein each disk drive comprises at least one data storage region, and wherein each row of the disk drive array comprises at least one row stripe including a different one of the data storage regions of each of the disk drives in the same row, and wherein each column of the disk drive array comprises at least one column stripe including a different one of the data storage regions of each of the disk drives in the same column, the method comprising:

• grouping data to be stored in a given row stripe during a write operation;
• calculating a row parity block for the given row stripe;
• writing the data and the associated row parity block to the corresponding disk drives of the array, thereby completing the write operation;
• calculating a column parity block for each column stripe modified during the write operation, wherein the calculating is performed after the completion of the write operation;
• and
• writing the column parity blocks to the disk drives storing the parity blocks for the modified column stripes.

A method for storing data within a plurality of disk drives logically arranged to form a two-dimensional disk drive array having a plurality of rows and columns, and wherein each disk drive comprises at least one data storage region, and wherein each row of the disk drive array comprises at least one row stripe including a different one of the data storage regions of the disk drives in the same row, and wherein each column of the disk drive array comprises at least one column stripe including a different one of the data storage regions of the disk drives in the same column, the method comprising:

• grouping data to be stored in a given row stripe during a write operation;
• calculating a column parity block for each column stripe modified during the write operation;
• writing the data and the column parity blocks to the corresponding disk drives of the array, thereby completing the write operation;
• calculating a row parity block for the given row stripe, wherein the calculating is performed after the completion of the write operation;
• and
• writing the row parity block to the disk drive storing the parity block for the given row stripe.

A method for repairing a disk drive array following failure of at least one of the disk drives, wherein the disk drive array comprises a plurality of disk drives logically arranged in multiple rows and multiple columns, wherein at least one disk drive in each row of the disk drive array stores row parity data, and wherein at least one disk drive in each column of the disk drive array stores column parity data, the method comprising:

• selecting a failed disk drive;
• determining if data stored in the selected failed disk drive is recoverable using either:
• (i) data stored in other disk drives in the row in which the selected failed disk drive resides, or (ii) data stored in other disk drives in the column in which the selected failed disk drive resides;
• performing the following steps if the data stored in the selected failed disk drive is recoverable:
• recovering the data stored in the selected failed disk drive;
• writing the recovered data to a spare disk drive;
• replacing the selected failed disk drive in the disk drive array with the spare disk drive;
• repeating the above steps until all failed disk drives have been replaced by spare disk drives;
• wherein if the selected failed disk drive resides in a given row of the disk drive array, the data stored in the selected failed disk drive is not recoverable using data stored in other disk drives in the given row if:
• (i) any of the other disk drives in the given row is not operational, or (ii) row parity data stored in any of the other disk drives in the given row is not current;
• and
• wherein if the selected failed disk drive resides in a given column of the disk drive array, and the data stored in the selected failed disk drive is not recoverable using data stored in other disk drives in the given column if:
• (i) any of the other disk drives in the given column is not operational, or (ii) column parity data stored in any of the other disk drives in the given column is not current.