Parallel computer system including parallel storage subsystem including facility for correction of data in the event of failure of a storage device in parallel storage subsystem

• US 5,987,622 A
• Filed: 12/10/1993
• Issued: 11/16/1999
• Est. Priority Date: 12/10/1993
• Status: Expired due to Term

• Current Assignee: TM Patents LP
• Original Assignee: TM Patents LP
• Inventors: Lee, Roger, Sharakan, Eric, Lo Verso, Susan, Rowe, Eric L.
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Elmore, Stephen C.
• Application Number:
  US08/166,345
• Time in Patent Office: 2,167 Days
• Field of Search: 395/575, 395/200, 714/6, 714/11, 714/766, 714/800, 714/805, 714/763, 714/764, 711/5, 364/243, 364/228.3
• US Class Current: 714/6.1
• CPC Class Codes: G06F 11/1008 in individual solid state devices G06F11/1004 takes precedence
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems

Abstract

A parallel computer system comprising a plurality of processing elements and an input/output system interconnected by a communications network. The input/output system includes a plurality of input/output devices connected to the communications network for receiving messages from, and transmitting messages to, the processing elements over the communications network. The processing element during the storage operation generates a series of messages for transmission over the communications network to successive input/output devices. Each message includes a data item from one of a series of storage locations. The processing element generates for each stripe of a selected number of data items a check value for transmission in a message to another of the input/output devices. During a reconstruction operation, the processing element receives the data items and the check values from at least some of the input/output devices and performs a reconstruction operation in connection with each check value and a stripe of data values for which the check value was generated during the storage operation thereby to reconstruct data items stored in at least one other of the input/output devices.

Claims

1. A parallel computer system comprising a plurality of processing elements and an input/output system interconnected by a communications network:

A. the input/output system including a plurality of input/output devices connected to said communications network for receiving messages from, and transmitting messages to, said processing elements over said communications network; and

B. each processing element including a memory including a series of storage locations for storing data items, the data items in the series of storage locations defining a series of stripes each including data items in a predetermined number of successive storage locations, the processors of all of said processing elements;

i. during a storage operation generating a series of messages for transmission over said communications network to successive input/output devices, each message including a data item from one of said storage locations, each processor generating for each stripe of data items a check value for transmission in a message to another of said input/output devices; and

ii. during a reconstruction operation, receiving said data items and said check values from at least some of said input/output devices and performing a reconstruction operation in connection with each check value and a stripe of data values for which the check value was generated during the storage operation thereby to reconstruct data items stored in at least one other of said input/output devices.

2. A parallel computer system as defined in claim 1 in which each said processor transmits a series of messages generated for each stripe of data items and the check value to different ones of said input/output devices.

3. A parallel computer system comprising a plurality of processing elements and an input/output system interconnected by a communications network:

A. the input/output system including a plurality of input/output devices defining a reconstruction group, at least some of said input/output devices during a reconstruction operation transmitting messages over said communications network to each processing element including a different data item from a stripe of data items or a check value associated therewith; and

B. each processing element, during a reconstruction operation, receiving data items and check values transmitted thereto by said input/output devices and performing a reconstruction operation in connection with each check value and a stripe of data values for which the check value was generated during a storage operation thereby to reconstruct data items in at least one other of said input/output devices.

4. A parallel computer system comprising a plurality of processing elements and an input/output system interconnected by a communications network:

A. the input/output system including a plurality of input/output devices connected to said communications network for receiving messages from, and transmitting messages to, said processing elements over said communications network; and

B. each processing element including a memory including a series of storage locations for storing data items, the data items in the series of storage locations defining a series of stripes each including data items in a predetermined number of successive storage locations, each said processing element during a storage operation generating a series of messages for transmission over said communications network to successive input/output devices in said input/output system, each message including a data item from one of said storage locations, the processing element generating for each stripe of data items a check value for transmission in a message to another of said input/output devices, thereby to store different data items in each stripe and the associated associated check value on diverse ones or the input/output devices.