Abstract: A database-management system (10) generates bounded-disorder indexes on its database keys. In such an index, the leaf nodes (51, 62) are large and are divided into a number of buckets (52, 54, 56, 58), only one of which ordinarily is accessed in any given single-record database operation. The key values in a leaf node are distributed among the leaf node's buckets in accordance with a hashing function. The lockable ranges locked for scanning functions are defined in accordance with key-valued locking, in which each lockable range is bounded by successive key values that exist in the database. But the multiple-bucket accesses that would otherwise be required, because of the hash-function distribution of key values among a node's several buckets, are avoided because the lockable ranges are defined by the sequence of key values in the bucket rather than in the node.

Abstract: Fast recovery from process terminations during atomic database transactions is accomplished by maintaining global memory after process terminations. Database data stored in global memory is accessed through control data structures. Control data structures are updated within atomic transactions that maintain structure consistency. Should a process fail during an atomic transaction that is updating control data structures, consistency is maintained by recovering the control data structures to their initial state existing immediately prior to the atomic transaction. Transactions involving control data structures are journaled as before-images in a log located in global memory. After a process failure, the before-images are installed to return the control data structures to the consistent state existing before the process failure.

Abstract: A two-phase commit protocol for a distributed transaction processing system employs the presumed-commit configuration, with the exception that the new presumed-commit protocol coordinator needs to force-write only a "commit" log record for committed transactions, not the previous force writing of two log records. In order to provide information needed to allow the coordinator to answer inquiries from subordinate processes following a crash or loss of communications, a technique for circumscribing the set of indeterminate transactions is employed. The transactions are numbered in increasing order, identified by a transaction ID (T.sub.-- ID). The commit protocol is not allowed to begin unless the transaction ID of the committing transaction is within some preselected range of numbers starting from the highest-numbered stably-recorded transaction ID.

Abstract: When a subtransaction (46) of a higher-level transaction (50) commits during the operation of a database (10), the database enters into its operation log a record (FIG. 6 ) that acts both as a commit record for the subtransaction (46) and as an update record for the higher-level transaction and includes a field (74) that identifies a higher-level "undo" transaction whereby the subtransaction can be undone without individually undoing its constituent operations. By logging operations in this manner, the database can handle multi-level recovery with very few restrictions on the timing of its updates and log entries.

Abstract: The present invention includes an approach to index tree structure changes which provides high concurrency while being usable with many recovery schemes and with many varieties of index trees. The present invention permits multiple concurrent structure changes. In addition, all update activity and structure change activity above the data level executes in short independent atomic actions which do not impede normal database activity. Only data node splitting executes in the context of a database transaction. This feature makes the approach usable with diverse recovery mechanisms, while only impacting concurrency in a modest way. Even this impact can be avoided by re-packaging the atomic actions, at the cost of requiring more from the recovery system.

Abstract: A distributed database system has a plurality of databases located at distinct nodes, at least one of the databases comprising a timestamping database. Distributed transactions are committed using a two phase protocol. During the first phase, each cohort to the transaction votes to commit or abort the transaction, and also votes an earliest time and a latest time at which the transaction is to be committed. If all the cohorts vote to commit the transaction and the intersection of the voted time ranges is not empty, then the transaction is committed during the second phase of the protocol. A transaction time is selected from the intersection of the voted time ranges and is used to timestamp all updated data that is durably stored when the transaction is committed. Before the first phase of the two phase commit protocol, each transaction read or write locks data at each node for which it needs read or write access.

Abstract: A key-accessed (indexed) file is organized such that the file structure consists only of two levels, an index level and a data level. Both levels are permanently stored on a page-organized secondary storage medium that supports random accessing of the pages. The index level is designed to have a fixed and specifiable number of pages and is stored entirely in the computer's memory when the file is in use. The fixed size of the index is made possible by having each index entry reference a data node with a growing (or shrinking) number of data pages as the file changes in size. Avoiding the accessing of more than one of the data pages referenced by an index entry is accomplished by means of an address computation that utilizes bits of the search argument.