Ensuring consistency in long running transactions

Abstract

Flow composition languages permit the construction of longrunning transactions from collections of independent atomic services. Due to environmental limitations, such transactions usually cannot be made to conform to standard ACID semantics. We propose set consistency, a powerful, yet intuitive, notion of consistency for long-running transactions. Set consistency considers the collection of permanent (nonintermittent) changes made by a process, when viewed at the end of its execution. Consistency requirements for such collections of changes are specified as predicates over the atomic actions of a process. Set consistency generalizes selfcancellation, a standard consistency requirement for longrunning transactions, where failed processes are responsible for undoing any partially completed work. Set consistency can also express strictly stronger requirements, such as mutual exclusion or dependency.We show that the set consistency verification problem for processes is co-NP complete and present an algorithm for verifying set consistency by reduction to propositional validity. We have implemented this algorithm and demonstrate the value and tractability of our approach on three real-world case studies. In each case, the consistency requirements can be verified within a second, demonstrating the practicality of our approach.

Recovery within long-running transactions

“…-Set Consistency: The set consistency approach [Fischer and Majumdar 2007] discards ordering and repetition of transaction actions and represents a transaction execution in terms of a set of successfully completed actions. Any actions which have either not been executed or have failed are not members of the set.…”

Research of cross-organizational services composition consistency using CSP trace semantic

“…We introduce an operator ‫ڇ‬ : ࣬ ൈ ࣬ ՜ ʹ ࣬ . For example, two runs p and q, p‫ڇ‬q produces the set of all interleaving of p and q [10].…”

“…Reference [10] gives an algorithm for verifying set consistency, they define a syntax-directed analysis which takes a process as input and constructs a predicate Ԅ whose satisfying assignments precisely represent the feasible execution set. Ԅ is composed from atomic predicates ȭ using logical operators ‫ר‬ǡ ‫ש‬ǡ , ื ุ.Then a predicate Ԅ over atomic predicates in ȭ defines a set of execution ࣟ(Ԅ) = {e ‫א‬ ʹ ஊ ȁ ٧ Ԅ }.…”

“…Ԅ is composed from atomic predicates ȭ using logical operators ‫ר‬ǡ ‫ש‬ǡ , ื ุ.Then a predicate Ԅ over atomic predicates in ȭ defines a set of execution ࣟ(Ԅ) = {e ‫א‬ ʹ ஊ ȁ ٧ Ԅ }. We expand the functions from seven to thirteen, ĭ Ȱ Ȱ Ǩ Ȱ ǫ ǡ Ȱ Ǩ Ȱ Ǩ Ȱ ԥ Ȱ Ǩǫ ǡ Ȱ , Ȱ ǫ , Ȱ ǫǨ , Ȱ ǫ , Ȱ ǫǫ , then it will solve more problem than [10] does. See details in Table 3.…”

“…But LTL is often overkill [13]. Reference [10] defines a simple notion of correctness for long running transaction, called set consistency. A set consistency specification captures the set of services which made observable changes, when viewed at the end of a process's execution.…”