Oracle International Corporation

Stefan Heinrich Roesch

Carol L Colrain

Kevin S Neel

Douglas N Surber

Nancy Reiko Ikeda

Angelo Pruscino

Wilson Wai Shun Chan

Michael Zoll

Tong Zhou

Paul Parkinson

Todd Jeffery Little

Jean De Lavarene

Tolga Yurek

Vivekanandhan Raja

Eugene Reddy Ho

Edward Heeren

Carol Lyndall Colrain

Nancy R Ikeda

Techniques are described for providing a guaranteed commit outcome for global transactions that are optimized by deferring the commit processing to a resource manager, such as a DBMS, using optimizations. The 'Distributed Transaction Processing: The XA Specification' standard specification is ambiguous regarding commit outcomes for transactions managed by DBMS, resulting in wrong results for vendors implementing the standard. The techniques described provide a guaranteed commit outcome when using the optimizations, creating the opportunity for safe replay of global transactions when a communication error or timeout occurs between the transaction manager and the resource manager, and eliminating ambiguous transaction outcomes reaching applications and end users. Techniques described herein distinguished between global transactions in a session managed by a transaction manager, and those in the same session that are managed by the resource manager using the optimizations. The techniques provide a guaranteed commit outcome when the commit is managed by the resource manager, or when a transaction manager is managing the transaction. Switching between the different techniques to provide a guaranteed outcome occurs in a safe, performing, and silent manner, based on who controls the current transaction in a session. The solution includes one-phase processing, read only optimizations, and promotable transactions.

Guaranteed commit outcome in a distributed transaction processing system

A system and method for reducing communications overhead in a distributed transaction processing environment such as an XA environment. In accordance with an embodiment communication overhead is reduced in a transaction by deferring a transaction end communication. The deferred transaction end communication is determined from a subsequent XA transaction communication thereby reducing the number of networks calls required for each transaction and enhancing system performance.

System and method for reducing communications overhead in a distributed transactions environment by modifying implementation of the transaction end function

A method, machine, and computer-readable medium is provided for managing transactional sets of commands sent from a client to a server for execution. A server receives, from a client in a second session, a request that identifies a set of commands that were sent by the client in a first session. The set of commands comprises one or more transactions. The server determines that at least one transaction of the one or more transactions in the set of commands has not committed. Based at least in part on determining that the at least one transaction of the one or more transactions has not committed, the server blocks, from the second session, completion of the at least one transaction in the first session to prevent committing of the at least one transaction in the first session.

Idempotence for database transactions

A system and method for reducing communications overhead in a distributed transaction processing environment such as an XA environment. In accordance with an embodiment communication overhead is reduced in a transaction start method by deferring a transaction start communication. The deferred transaction start communication is determined from a subsequent transaction processing communication thereby reducing the number of networks calls required for each transaction and enhancing system performance.

System and method for reducing communications overhead in a distributed transactions environment by modifying implementation of the transaction start function

A process, apparatus, and computer-readable medium are provided for rebuilding a database session when a previous database session becomes unavailable and the commands previously sent for execution on the previous database session satisfy certain criteria. The process includes determining whether or not a set of commands sent by a client for execution on the previous database session is acceptable to replay based at least in part on whether or not the set of commands satisfies one or more criteria. The process further includes determining that the previous database session is unavailable due to a planned or unplanned recoverable error. In response to determining that the previous database session is unavailable, if the set of commands is acceptable for replay, the set of commands is sent for execution on a new database session to rebuild the state, which was exposed to the client from the previous database session, on the new database session. The process masks the outage from the application.

Recovering stateful read-only database sessions

Systems and methods are provided for supporting transaction recovery based on inline delegation of transaction logs to a database. The system can retrieve transaction recovery information of a transaction from a persistence store, wherein the transaction recovery information is persisted on a resource manager that is associated with the persistence store during a prepare phase. Furthermore, the system can retrieve one or more in-doubt transactions from one or more participating resource managers of the transaction. Then, the system can determine whether to commit or roll back the one or more in-doubt transactions by matching the one or more in-doubt transactions to the transaction recovery information.

Systems and methods for supporting inline delegation of middle-tier transaction logs to database

Transactional and non-transactional data for maintaining session state is described. In a session, a server instance receives, from a client, a request to execute transactional commands and non-transactional commands. The server instance sends at least one first directive executable at the client to store, in a first set of information, one or more first items representing one or more particular transactional commands that are open but not yet committed. The first set of information is maintained separately from a second set of information including one or more second items representing one or more particular non-transactional commands that have been processed. Based on committing one or more particular transactional commands, the server instance sends at least one second directive executable at the client to purge, from the first set of information, the one or more first items without purging, from the second set of information, the one or more second items.

Transactional and non-transactional data for maintaining session state

A method, machine, and computer-readable medium is provided for managing transactional sets of commands sent from a client to a server for execution. A first server reports logical identifiers that identify transactional sets of commands to a client. The first server commits information about a set of commands to indicate that the set has committed. A second server receives, from the client, a request that identifies the set based on the logical identifier that the client had received. The second server determines whether the request identified the latest set received for execution in a corresponding session and whether any transactions in the set have not committed. If any transaction has not committed, the second server enforces uncommitted state of the identified set by blocking completion of the identified set issued in the first session. The identified set may then be executed in the second session without risk of duplication.

Methods, devices, and computer-readable media are provided for restoring state that was built up on a first session between a first server instance and a client to a second session between a second server instance and the client. Non-transactional session state that existed for the first session is preserved by repeating non-transactional commands in the second session. Transactions are executed in the second session when the transactions did not complete in the first session. The first server instance sends, to the client in the first session, information to maintain for a possible replay of commands that were sent in a request to the first server instance for execution in the first session. If the first session becomes unavailable, the maintained information may be used by the second server instance to restore the database session, masking the outage from users, applications, and clients.

Masking database outages from clients and applications

A process, apparatus, and computer-readable medium are provided for rebuilding a database session when a previous database session becomes unavailable and the commands previously sent for execution on the previous database session satisfy certain criteria. The process includes determining whether or not a set of commands sent by a client for execution on the previous database session is acceptable to replay based at least in part on whether or not the set of commands satisfies one or more criteria. The process further includes determining that the previous database session is unavailable. In response to determining that the previous database session is unavailable, if the set of commands is acceptable for replay, the set of commands is sent for execution on a new database session to rebuild the state on the new database session. The process masks the outage from the application.

Techniques are provided for responding to the termination of a node by selecting another node, and assigning to the selected node the affinity relationships that existed between the terminated node and one or more objects. The resources that belong to the objects involved in the affinity relationships are remastered to the selected node. The selected node then performs recovery of the resources that had been opened by the terminated node and/or serves as a failover node to execute the transactions that had been executing on the terminated node.

Affinity-based recovery/failover in a cluster environment

Techniques are provided for responding to the termination of a node executing one or more transactions by selecting another node to complete the transactions, and assigning to the selected node the affinity relationships that existed between the terminated node and the objects being accessed by said transactions.

Optimization for transaction failover in a multi-node system environment where objects' mastership is based on access patterns

Techniques are provided for remastering resources based on node utilization. According to one such technique, resources are remastered in response to the over-utilization of the node that currently masters those resources. The utilization of each node is tracked, and when a particular node's utilization exceeds a specified threshold, selected resources that are currently mastered by that node are remastered so that nodes other than the particular node become the new masters for the selected resources. Each node's utilization is based on that node's capacity, and each node's capacity may differ. According to an aspect of one technique, each node's capacity is based on that node's processing resources and memory resources. Remastering resources in this manner tends to reduce the average amount of time taken for nodes to handle requests for the resources that they master.

Growing community of inventors

Redwood Shores, CA, United States of America

Stefan Heinrich Roesch