International Business Machines Corporation

Steven D Gerdt

James T Brady

Alden B Johnson

Norman Apperley

Dung Kim Nguyen

Lloyd R Shipman, Jr

Barrie N Harding

M Jaishankar Menon

Dennis R Escola

Jaishankar Moothedath Menon

John Richard Paveza

Juan Li

Hai-Fang Yun

Paul W Hunter

Raymond Eugene Garcia

Dung Lim Nguyen

Paul T Burton

Aspects for preventing deadlock in a log structured array are described. In an exemplary method aspect, and system for providing same, the method includes monitoring available segments to determine when to reserve an open segment, identifying a process type requesting access to the open segment, and limiting access to one process type for the reserved segment. Limiting access further includes limiting access to a garbage collection process.

Method and system for preventing deadlock in a log structured array

A computing system that incorporates the invention includes a host processor which is coupled to a memory subsystem via a first bus system, a controller device and a second bus system. The controller device includes memory for storing plural Scripts for replay to the host processor, for instance, via the second bus system. A Script is an instruction set used to execute operations on a controller device. Each Script includes one or more addresses where either message or status data (or other operational data) can be found which is to be inserted, prior to dispatch of the Script. During operation of the computing system, the memory subsystem is caused, as a result of its operation, to issue an instruction to the controller device to dispatch a Script to, for instance, the host processor. The controller device responds by accessing the required Script, playing the Script which results in accesses to locally stored operational data for inclusion into the Script. Local storage of this data avoids unnecessary data transfers over the bus system to the memory subsystem to obtain the required data for inclusion into a Script.

Increasing I/O performance through storage of packetized operational

A method for rebuilding contents of a malfunctioned direct access storage device within a log-structured array is disclosed. In accordance with the method and system of the present invention, each direct access storage device within a log-structured array is divided into multiple segment-columns, and each corresponding segment-column from each direct access storage device within the log-structured array forms a segment. A segment is first located within the direct access storage devices. A determination is made as to whether or not the segment is empty. In response to a determination that the segment is empty, a pointer is moved within a segment-column mapping table from pointing to a segment-column in the malfunctioned direct access storage device to point to a segment-column in a spare direct access storage device of the segment. In response to a determination that the segment is not empty, rebuilding contents of the segment-column in the malfunctioned direct access storage device to the segment-column in the spare direct access storage device, and moving the pointer within the segment-column mapping table from pointing to the segment-column in the malfunctioned direct access storage device to point to the segment-column in the spare direct access storage device of the segment. The process then returns to the determination step until all segment-columns within the malfunctioned direct access storage device are rebuilt.

Method and system for rebuilding log-structured arrays

Predictive failure analysis of a storage subsystem is efficiently conducted and data quickly recovered from a failed Read operation. This may be implemented in a storage system including a host coupled to a supervising processor that couples to a parity-equipped RAID storage subsystem having multiple HDAs each including an HDA controller and at least one storage medium. In one embodiment, when an HDA experiences an error during a Read attempt, the HDA transmits a recovery alert signal to the supervising processor; then, the processor and HDA begin remote and local recovery processes in parallel. The first process to complete provides the data to the host, and the second process is aborted. In another embodiment, an HDA's PFA operations are restricted to idle times of the HDA. A different embodiment limits HDA performance of PFA to times when the processor is conducting data reconstruction. Another embodiment monitors HDA errors at the supervisory processor level, initiating an HDA's PFA operations when errors at that HDA have a certain characteristic, such as a predetermined frequency of occurrence.

Efficient system for predicting and processing storage subsystem failure

A system for managing data storage devices of a data storage subsystem. A data storage system includes a controller coupled to multiple data storage devices. In response to a request, the controller allocates the devices' storage space into a number of storage partitions or 'virtual devices.' The request specifies the size and function mode for each storage partition. Each storage partition, having the requested size, is operated according to its requested function mode. This involves mapping access commands, which specify virtual addresses, to the proper physical addresses of the appropriate data storage devices.

Reconfigurable data storage system

A data storage and retrieval system is described having one or more key-lock storage areas which can be modified under the control of commands received from an attached computer. Each key-lock is stored in nonvolatile data storage. A key-lock contains a key field and a lock field. The contents of the key-lock are read and modified based on commands received from the computer. The command to write a new key-lock value specifies a first key and a proffered key-lock. The subsystem reads the current key-lock from nonvolatile storage, then compares the first and current keys. When the first and current keys are equal, the proffered key-lock is stored in the key-lock storage area of nonvolatile data storage, thereby replacing the original key-lock. When the keys are not equal, the key-lock is left unmodified and status data is transmitted to the computer indicating that the update request has failed. The subsystem will send the original key-lock value to the computer upon demand. One alternative embodiment of the invention has means for designating a predetermined unlocked key value as indicative of the key-lock being unlocked.

Storage subsystem having a modifiable key-lock

A data storage system is described which provides for fault tolerance during execution of a media maintenance program on a selected track of nonvolatile data storage. Checkpoint data for the media maintenance program is stored in nonvolatile storage prior to writing control information on the selected track during the execution of the media maintenance program. The checkpoint data includes an address in the media maintenance program for resuming execution in the event that the media maintenance process is aborted. The checkpoint data is read from the nonvolatile storage area after media maintenance has been aborted by a fault, typically after power has been reapplied to the system. The execution environment for the media maintenance program is restored by storing data in the RAM of the controller which duplicates the control data needed by the particular segment of the media maintenance program which will be restarted. After the environment is restored, the execution of the media maintenance program resumes at the address in checkpoint data stored prior to writing the control data on the selected track.

Growing community of inventors

San Jose, CA, United States of America

Steven D Gerdt