Nested transactions and read-write locking
PODS '87 Proceedings of the sixth ACM SIGACT-SIGMOD-SIGART symposium on Principles of database systems
Timestamp-Based Orphan Elimination
IEEE Transactions on Software Engineering
On the correctness of orphan management algorithms
Journal of the ACM (JACM)
Hybrid concurrency control for abstract data types
Proceedings of the seventh ACM SIGACT-SIGMOD-SIGART symposium on Principles of database systems
| Hi-index | 0.00 |
This thesis defines a property called "view-serializability", which formalizes internal consistency for a system of nested atomic transactions. Internal consistency is a stronger condition than the usual notion of data base consistency, because it takes into account the views of transactions which will never commit. In a distributed system, local aborts of remote subactions and crashes of nodes can generate orphans: active actions which are descendants of actions that have aborted or are guaranteed to abort. Because it is not always feasible or efficient to eliminate orphans immediately, special care is needed to insure that they see consistent system states when they are allowed to continue running. We investigate a particular dynamic detection strategy designed to detect orphans before they violate internal consistency. This algorithm piggybacks abort and crash information on the normal messages between nodes. We consider a simpler algorithm that only handles orphans arising from explicit aborts. We describe the simplified orphan detection algorithm at various levels of abstraction, using an algebraic model convenient for describing asynchronous systems. The highest-level model is specified in terms of a (virtual) global state. At this level of abstraction we require that the states generated by the model satisfy view-serialability. Lower-level models progressively localize the description of the algorithm''s operation, and the lowest level of abstraction presents a fully distributed model of the (simplified) orphan detection scheme.