US5488731A — Synchronization method for loosely coupled arrays…

Claim 1

Independent Claim 225 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, nonvolatile random access memory (NVRAM) and a processor, said multiprocessor system storing RAID-structured data across disk drive means in a plurality of different nodes, said system performing a method comprising the steps of: a. listing at least an identifier of a data segment to be updated by received update data in an NVRAM in a first node in response to a command to write said update data to said data segment; b. sending said update data from said first node to a second node containing a copy of said data segment; c. removing said listing of said identifier in said NVRAM in said first node only when said update data is written to disk drive means in said first node and after receiving a signal that said second node has recorded said update data; d. causing each node, in the event of a power-up, to scan its NVRAM to find any listed identifiers of data segments contained therein; and e. for any data segment denoted by a listed identifier in said NVRAM in said first node, causing a corresponding data segment in said second node to be in synchronism with said data segment denoted by said listed identifier in NVRAM in said first node.

Claim 2

Dependent Claim 225 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, nonvolatile random access memory (NVRAM) and a processor, said multiprocessor system storing RAID-structured data across disk drive means in a plurality of different nodes, said system performing a method comprising the steps of: a. listing at least an identifier of a data segment to be updated by received update data in an NVRAM in a first node in response to a command to write said update data to said data segment; b. sending said update data from said first node to a second node containing a copy of said data segment; c. removing said listing of said identifier in said NVRAM in said first node only when said update data is written to disk drive means in said first node and after receiving a signal that said second node has recorded said update data; d. causing each node, in the event of a power-up, to scan its NVRAM to find any listed identifiers of data segments contained therein; and e. for any data segment denoted by a listed identifier in said NVRAM in said first node, causing a corresponding data segment in said second node to be in synchronism with said data segment denoted by said listed identifier in NVRAM in said first node.

Claim 3

Dependent Claim 46 words

The method as recited in claim 1 wherein step e uses said update data to enable an update action to be accomplished in said second node for each data segment that corresponds to an identifier of a data segment listed in NVRAM in said first node.

Claim 4

Dependent Claim 21 words

The method as recited in claim 1 wherein step e enables said disk drive means to store identical non-updated data segments.

Claim 5

Dependent Claim 39 words

The method as recited in claim 1 wherein each said data segment is a data block having a block identifier, said block identifier being listed in a said NVRAM if said data segment is in process of being updated.

Claim 6

Dependent Claim 38 words

The method as recited in claim 1 wherein step a, in addition to listing the identifier of said data segment in said NVRAM in first node, also lists said update data in said NVRAM in said first node.

Claim 7

Dependent Claim 26 words

The method as recited in claim 1 wherein step c only occurs after said second node has written said update data to its disk drive means.

Claim 8

Dependent Claim 201 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, a nonvolatile random access memory (NVRAM) and a processor, said multiprocessor system storing RAID-structured data across disk drive means in a plurality of different nodes, said system performing a method comprising the steps of: responding to a command to write new data to replace old data in a data segment stored in a first node, by storing in NVRAM in a different node which stores parity data corresponding to old data stored in said first node, an entry comprising said new data, a state indication, and a copy of new parity as calculated based upon an exclusive-or combination of old data from said first node, old parity from said different node and said new data; transmitting said new data to said first node for storage therein, and upon receiving a signal acknowledging successful storage, causing said different-node to reset said state indication; and causing said different node, in the event of a power-up to scan its NVRAM for a said entry, and upon finding a said entry with a non-reset state indication, transmitting said new data to said first node.

Claim 9

Dependent Claim 201 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, a nonvolatile random access memory (NVRAM) and a processor, said multiprocessor system storing RAID-structured data across disk drive means in a plurality of different nodes, said system performing a method comprising the steps of: responding to a command to write new data to replace old data in a data segment stored in a first node, by storing in NVRAM in a different node which stores parity data corresponding to old data stored in said first node, an entry comprising said new data, a state indication, and a copy of new parity as calculated based upon an exclusive-or combination of old data from said first node, old parity from said different node and said new data; transmitting said new data to said first node for storage therein, and upon receiving a signal acknowledging successful storage, causing said different-node to reset said state indication; and causing said different node, in the event of a power-up to scan its NVRAM for a said entry, and upon finding a said entry with a non-reset state indication, transmitting said new data to said first node.

Claim 10

Dependent Claim 44 words

The method as recited in claim 7 wherein a said state indication indicates an unsynchronized state, said entry including a further state indication which, when set, indicates that new parity has not yet been written to disk in said different node's disk drive means.

Claim 11

Dependent Claim 28 words

The method as recited in claim 8 wherein said data segment is a block of data and a said entry includes an identifier of said block of data.

Claim 12

Dependent Claim 44 words

The method as recited in claim 9 wherein said different node, upon scanning its NVRAM at power-up, first accesses an entry having oldest received new data for said block of data in said first node and proceeds to process subsequent entries in time order.

Claim 13

Dependent Claim 31 words

The method as recited in claim 10 wherein said different node, upon receiving an acknowledgement of a successful storage of new data by said first node, resets said unsynchronized state indicator.

Claim 14

Dependent Claim 224 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, nonvolatile random access memory (NVRAM) and a processor, said multiprocessor'system storing RAID-structured data across a disk drive means in a plurality of different nodes, said system comprising: means for listing at least an identifier of a data segment to be updated by update data in an NVRAM in a first node in response to a command to write said update data to said data segment in said first node; means for sending said update data from said first node to a second node containing a copy of said data segment; means for removing said listing of said data segment in said NVRAM in said first node only when said update data is written to disk drive means in said first node and after receiving a signal that said second node has recorded said update data; means for causing each node, in the event of a power-up, to scan its NVRAM to find any listed identifiers of data segments contained therein; and means, responsive to finding a data segment identifier listed in said NVRAM in said first node, for causing a corresponding data segment in said second node to be in synchronism with said data segment listed in said NVRAM in said first node.

Claim 15

Dependent Claim 224 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, nonvolatile random access memory (NVRAM) and a processor, said multiprocessor'system storing RAID-structured data across a disk drive means in a plurality of different nodes, said system comprising: means for listing at least an identifier of a data segment to be updated by update data in an NVRAM in a first node in response to a command to write said update data to said data segment in said first node; means for sending said update data from said first node to a second node containing a copy of said data segment; means for removing said listing of said data segment in said NVRAM in said first node only when said update data is written to disk drive means in said first node and after receiving a signal that said second node has recorded said update data; means for causing each node, in the event of a power-up, to scan its NVRAM to find any listed identifiers of data segments contained therein; and means, responsive to finding a data segment identifier listed in said NVRAM in said first node, for causing a corresponding data segment in said second node to be in synchronism with said data segment listed in said NVRAM in said first node.

Claim 16

Dependent Claim 202 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, a nonvolatile random access memory (NVRAM) and a processor, said multiprocessor system storing RAID-structured data across disk drive means in a plurality of different nodes, said system comprising: means for responding to a command to write new data to replace old data in a data segment stored in a first node, by storing in NVRAM in a parity node which stores parity data corresponding to data stored in said first node, an entry comprising said new data, a state indication, and a copy of new parity as calculated based upon an exclusive-or combination of old data from said first node, old parity from said parity node and said new data; means for transmitting said new data to said first node for storage therein, and upon receiving a signal acknowledging successful storage, causing said parity node to reset said state indication; and means for causing said parity node, in the event of a power-up to scan its NVRAM for a said entry, and upon finding a said entry with a non-reset state indication, to transmit said new data to said first node.

Claim 17

Dependent Claim 202 words

A multiprocessor system including a plurality of substantially identical nodes interconnected through a switch network, each node comprising disk drive means, a nonvolatile random access memory (NVRAM) and a processor, said multiprocessor system storing RAID-structured data across disk drive means in a plurality of different nodes, said system comprising: means for responding to a command to write new data to replace old data in a data segment stored in a first node, by storing in NVRAM in a parity node which stores parity data corresponding to data stored in said first node, an entry comprising said new data, a state indication, and a copy of new parity as calculated based upon an exclusive-or combination of old data from said first node, old parity from said parity node and said new data; means for transmitting said new data to said first node for storage therein, and upon receiving a signal acknowledging successful storage, causing said parity node to reset said state indication; and means for causing said parity node, in the event of a power-up to scan its NVRAM for a said entry, and upon finding a said entry with a non-reset state indication, to transmit said new data to said first node.

Annuity Stage	Window Opens	Payment Deadline	Surcharge Period Until	Estimated Fee	Status
3.5 Years (Year 4)	1999-01-29	1999-07-31	2000-01-30	$2000.00	Window Closed / Lapsed
7.5 Years (Year 8)	2003-01-29	2003-07-31	2004-01-30	$3760.00	Window Closed / Lapsed
11.5 Years (Year 12)	2007-01-29	2007-07-31	2008-01-30	$7700.00	Window Closed / Lapsed

Synchronization method for loosely coupled arrays of redundant disk drives

Abstract

Inventors

Bibliographic Details

Claims Analysis

References Cited (Backward Citations)

USPTO Annuity & Maintenance Windows

Assigned Categories & Tags

Assign Category Tag