Abstract: Methods and apparatus for the rapid generation of a snapshot copy of the data stored in a RAID storage subsystem. In addition to the users configured RAID logical units, the present invention provides for the definition within the RAID controller of a logical RAID level one device having an operational half comprising the users defined logical unit(s) and having a non-operational mirror component. The user access data stored on the RAID subsystem by direct access to the users defined logical units. When a user directs a snapshot copy request to the operational, user defined logical units, the RAID controller responds by temporarily configuring available storage capacity (e.g. spare disk drives) in the RAID subsystem to perform the function of the non-operational mirror component of the logical RAID level one device.

Abstract: Methods and apparatus for moving pinned data corresponding to a temporarily dead RAID device between the cache memory of a RAID subsystem and a log area. In response to detection of a dead RAID device within a RAID subsystem, the methods of the present invention move any pinned data from the cache memory of the RAID controller to a log area preferably allocated on the disk space of one or more operational RAID devices within the subsystem. In response to revival of the dead RAID device methods of the present invention restore the logged, pinned data from the log area of the operational RAID device(s) to the cache memory as dirty data ready for posting to the revived RAID device. The log area may be either permanently allocated within the RAID subsystem, or may be dynamically allocated in response to recognition of the dead RAID device and freed in response to revival of the RAID device.

Abstract: Methods and associated data structures operable in a RAID subsystem to improve I/O performance. A two level cache data structure and associated methods are implemented with a RAID controller. The lower level cache comprises buffers holding recently utilized blocks of the disk devices. The upper level cache records which blocks are present in the lower level cache for each stripe in the RAID level 5 configuration. The upper level cache serves to reduce the overhead processing required of the RAID controller to determine which blocks are present in the lower level cache. Having more rapid access to this information by lowering the processing overhead enables the present invention to rapidly select between different write techniques to post data and error blocks from low level cache to the disk array. A RMW write technique is used to post data and error checking blocks to disk when insufficient information reside in the lower level cache.

Abstract: Methods and associated apparatus operable in a RAID subsystem having redundant disk array controllers to assure dual, redundant cache consistency while permitting operation of the RAID subsystem in response to host computer I/O requests. The methods and apparatus of the present invention provide for a process of initializing the redundant disk array controllers in response to a power-on reset cycle (or other similar reset cycles). The first controller partially initializes for processing of host requests then awaits the partial initialization of the second controller. After a brief timeout awaiting the second controller, or in response to an erroneous initialization of the second controller, the first controller configures itself to perform host computer I/O requests in a write-through mode (bypassing cache operations) until the second controller eventually initializes. Once both controller are initialized, the redundant caches are synchronized.

Abstract: Method for implementing a stripe write operation in a RAID device having XOR command set enabled disk drives on a common interface bus. The method of the present invention improves upon prior designs which do not use the XOR command set by eliminating the need for the RAID controller to include XOR (parity) generation computational elements. Further, the method of the present invention improves upon the stripe write method suggested by the XOR command set specifications in which a stripe is performed by a series of xdwrite XOR command operations issued to the data drives of the RAID array. Rather, the method of the present invention performs parallel standard writes of the data portions of the stripe write, then issues a rebuild XOR command to the parity disk drive to rapidly regenerate the parity blocks in the stripe just written. The method of the present invention reduces the worst case rotational latency delay of the stripe write operation to two rotational latency periods.

Abstract: A high performance scaleable hardware architecture for a disk array storage subsystem which supports RAID modes 0, 3, 4 and 5. The architecture features a high bandwidth parity calculation engine, a buffered PCI interface operating at the full speed of the PCI bus, and a dedicated local memory. The dedicated local memory is dual ported so that PCI and parity operations may operate concurrently. The architecture of the disk array controller allows parity calculations and memory block moves to occur without interfering with the controller processor or its associated memory, freeing the controller processor to manage array task control. The array controller configuration allows simultaneous operation of data block moves between storage I/O devices and local memory; data block moves between host SCSI connections and local memory; parity calculations; and normal CPU memory fetches, queued operations for block moves and queued operations for parity tasks.

Abstract: A method of insuring recovery of a quantity write request data in a storage device having a plurality of controllers includes the steps of (1) providing a primary controller for storing the write request data received from a host device, (2) providing an alternate controller having a memory area allocated for use by the primary controller, (3) maintaining a first control block in the primary controller for indicating status of the write request data stored in the primary controller, and (4) maintaining a second control block in the alternate controller for indicating status of the write request data stored in the primary controller.