Maintaining Consistency Properties of Grid Workflows in Collaborative Editing Systems

Abstract

Real-time collaborative editing systems allow a group of users to view and edit the same item at the same time from geographically dispersed sites. Consistency maintenance in the face of concurrent accesses to shared entities is one of the core issues in the design of these systems. The paper introduces a locking based approach that grid application developer environments can apply to protect the consistency properties of workflows during collaborative editing scenarios. We developed two pessimistic locking algorithms to assure that under no circumstances can application developers break the acyclic criteria of workflow graphs or add invalid edges to them. We prove that the algorithms result consistent graphs, moreover, do not result the cancellation of any user's editing transaction. We see that the second, improved algorithm gives much better performance in terms of number of concurrent users that can work on a graph.

Efficient Graph Partitioning Algorithms for Collaborative Grid Workflow Developer Environments

“…no more than one user can have U lock on a component. On the other hand, we proved that multiple S locks can exists on the same component, moreover, in some cases both S lock and U lock is allowed on the same component [20]. It has been also proven that a lock request must be denied only, if it puts U lock on a component that already has S lock and at the same time it puts S lock on a component that already has U lock.…”

“…Our research focuses on the graph-level consistency rules because in a multi-user editing scenario no party has the complete view of the whole workflow, thus maintaining graph-level consistency is not self-evident. In [20] we defined a protocol that -if implemented by a lock evaluator algorithm -can prevent dangling edges and multiple incoming edges. The protocol basically prohibits users to hold lock on an edge without holding locks on the source and the sink vertexes of this edge.…”

“…In has been also shown in [20] that not a single, but several different protocols exist that can prevent cycles in workflows. These protocols partition workflow graphs in such a way that no edge creation operations within these sub-graphs can result cycle in the complete graph.…”

“…These protocols partition workflow graphs in such a way that no edge creation operations within these sub-graphs can result cycle in the complete graph. In [20] we defined two of such algorithms, and in [13] we gave a third algorithm. In the current paper our goal is to evaluate the usability of these algorithms, so we shortly summarize how the three algorithms work.…”

“…Recent works from this field studied the concurrent access to grid workflows [6] [11], but did not deal with inconsistency, a common issue in computer supported collaborative work (CSCW) environments. Our previous papers [13] [20] proposed lock based concurrency control and graph partitioning algorithms to protect the following three consistency criteria of workflows during collaborative editing sessions: (1) workflows must remain acyclic; (2) edges in the workflow must point to existing vertexes; (3) maximum one input edge can refer to an input parameter of a workflow node (maximum one edge can provide data for an input channel of a grid service/job). Those papers introduced three different graph partitioning algorithms that can allocate sub-graphs to workflow editors in such a way, that the modifications performed on these sub-graphs cannot break the consistency rules of the complete graph.…”