International Business Machines Corporation

Richard L Mattson

Jaishankar Moothedath Menon

Spencer W Ng

Henry Martin Gladney

Douglas J Lorch

Miguel M Blaum

Hsieh T Hao

Juan A Rodriguez-Rosell

A method for managing a cache hierarchy having a fixed total storage capacity is disclosed. The cache hierarchy is logically partitioned to form a least recently used (LRU) global cache and a plurality of LRU destaging local caches. The global cache stores objects of all types and maintains them in LRU order. In contrast, each local cache is bound to objects having a unique data type T(i), where i is indicative of a DataType. Read and write accesses by referencing processors or central processing units (CPU's) are made to the global cache. Data not available in the global cache is staged thereto either from one of the local caches or from external storage. When a cache full condition is reached, placement of the most recently used (MRU) data element to the top of the global cache results in an LRU data element of type T(i) being destaged from the global cache to a corresponding one of the local caches storing type T(i) data. Likewise, when a cache full condition is reached in any one or more of the local caches, the local caches in turn will destage their LRU data elements to external storage. The parameters defining the partitions are externally supplied.

Method for managing a cache hierarchy having a least recently used (LRU)

A controller for a disk array with parity and sparing includes a non-volatile cache memory and optimizes the destaging process for blocks from the cache memory to both maximize the cache hit ratio and minimize disk utilization. The invention provides a method for organizing the disk array into segments and dividing the cache memory into groups in order of least recently used memory locations and then determining metrics that permit the disk array controller to identify the cache memory locations having the most dirty blocks by segment and group and to identify the utilization rates of the disks. These characteristics are considered to determine when, what, and how to destage. For example, in terms of maximizing the cache hit ratio, when the percentage of dirty blocks in a particular group of the cache memory locations reaches a predetermined level, destaging is begun. The destaging operation continues until the percentage of dirty blocks decreases to a predetermined level. In terms of minimizing disk utilization, all of the dirty blocks in a segment having the most dirty blocks in a group are destaged.

Destaging modified data blocks from cache memory

A method and means is disclosed for dynamically partitioning an LRU cache partitioned into a global cache storing referenced objects of k different data types and k local caches storing objects of a single type. Referenced objects are stored in the MRU position of the global cache and overflow is managed by destaging the LRU object from the global to the local cache having the same data type. Dynamic partitioning is accomplished by recursively creating and maintaining from a trace of objects an LRU list of referenced objects and associated data structures for each subcache, creating and maintaining a multi-planar array of partition distribution data from the lists and the trace as a collection of all possible of maximum and minimum subcache sizing, optimally resizing the subcache partitions by applying a dynamic programming heuristic to the multiplanar array, and readjusting the partitions accordingly.

Method and means for dynamically partitioning cache into a global and

A method and system for minimizing seek affinity and enhancing write sensitivity in a direct access storage device (DASD) array are disclosed. SEEK affinity and WRITE efficiency are preserved in which logical cylinders, as recorded on the DASD array, are managed as individual log structured files (LSF). Tracks or segments of data and parity blocks having the same or different parity group affinity and stored on the same or different DASD cylindrical addresses are written into a directory managed buffer. Blocks having the same parity affinity and written to counterpart cylinders are written out from the buffer to spare space reserved as part of each DASD cylinder. Otherwise, blocks are updated in place in their DASD array location.

Method and system for minimizing seek affinity and enhancing write

A method and means for encoding data written onto an array of M synchronous DASDs and for rebuilding onto spare DASD array capacity when up to two array DASD fail. Data is mapped into the DASD array using an (M-1)*M data array as the storage model where M is a prime number. Pairs of simple parities are recursively encoded over data in respective diagonal major and intersecting row major order array directions. The encoding traverse covering a topologically cylindrical path. Rebuilding data upon unavailability of no more than two DASDs merely requires accessing the data array and repeating the encoding step where the diagonals are oppositely sloped and writing the rebuilt array back to onto M DASDs inclusive of the spare capacity.

Method and means for encoding and rebuilding data contents of up to two

A method and apparatus are disclosed which minimize the additional read and write loading for each remaining DASD in a DASD array occasioned by a single point of failure where subsets of DASDs may accessed asynchronously (RAID 4 or RAID 5). In this method, parity groups are written across the array in a pattern forming a balanced incomplete block design or the equivalent. In the event of failure of a DASD in the array, then any additional loading resulting from the failure is uniformly spread among the remaining DASDs.

Method and means for managing DASD array accesses when operating in

A method and means in which data and parity blocks forming parity groups together with spare blocks are distributed over array block locations according to at least one combinatorial design, each group having N data and P parity blocks. The combinatorial designs yield uniform or balanced loading thereby minimizing the number of accesses to reconstruct missing data and parity blocks and their copyback into spare block locations, and, minimize the number of accesses to the reconstructed data referenced subsequent to its copyback. Distributions of the spare block capacity of one or two DASDs are shown over single and multiple arrays and shared among multiple arrays. Parity block distribution although ancillary to spare distribution enhances throughput and reduces the number of accesses for rebuild etc.

Method and means for distributed sparing in DASD arrays

In a distributed processing system network in which at least one node operates as a source location having access to data objects of a database, and at least one other node operates as a replica location storing replicas of data objects from the source location, managing obsolescence of the replicas is performed by having the replica locations submitting requests to the source location for ascertaining obsolescence of data objects. The source location, responsive to a request from a requesting replica location, extracts identifiers of a set of obsolete objects and communicates them to the requesting replica location. Upon receiving the identifiers, the requesting location renders inaccessible those data objects corresponding to the identifiers received. The source location then removes those identifiers that have been communicated to the requesting replica location.

Communication for version management in a distributed information service

In a distributed processing system having a source node accessing data objects from a database and a replica node storing replicas of requested source data objects received from the source location, the impact to replicas caused by a change in a source data object is calculated by assigning a version number to the change. An identifier of the portion of the source database affected by the change is generated, as well as a list of replicas containing objects from the affected portion of the database. For a replica location communicating with the source location, a table of the replicas from the list is then recorded along with the version number for communicating to the replica location.

Impact calculation for version management in a distributed information

Short traces of consecutive CPU references to storage are accumulated and processed to ascertain hit ratio as a function of cache size. From this determination, an allocation of cache can be made. Because this determination requires minimal processing time, LRU-referenceable memory space among concurrently executing sequential processes is used dynamically by a CPU cache manager.

Growing community of inventors

San Jose, CA, United States of America

Richard L Mattson