Asynchronous Complete Distributed Garbage Collection

Veiga, Luís; Ferreira, Paulo

doi:10.1109/ipdps.2005.113

Cited by 13 publications

(16 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This design choice was influenced by an expected feature of compensation handlers: the possibility to compensate transactions partially executed or completed transactions, where by completed transaction we mean a transaction with no more active inputs, excluding transaction identifiers. This feature would not have problems of scalability in a real system, since it could be associated to our calculus a distributed garbage collection [19] to remove compensations of transactions no longer reachable. In Appendix A we provide partial reductions for two executions of the example described in Section 3.…”

Section: Operational Semanticsmentioning

confidence: 99%

Dynamic Recovering of Long Running Transactions

Vaz

Ferreira

Ravara

2009

Trustworthy Global Computing

View full text Add to dashboard Cite

Abstract. Most business applications rely on the notion of long running transaction as a fundamental building block. This paper presents a calculus for modelling long running transactions within the framework of the π-calculus, with support for compensation as a recovery mechanism. The underlying model of this calculus is the asynchronous polyadic π-calculus, with transaction scopes and dynamic installation of compensation processes. We add to the framework a type system which guarantees that transactions are unequivocally identified, ensuring that upon a failure the correct compensation process is invoked. Moreover, the operational semantics of the calculus ensures both installation and activation of the compensation of a transaction.

show abstract

Section: Operational Semanticsmentioning

confidence: 99%

Dynamic Recovering of Long Running Transactions

Vaz

Ferreira

Ravara

2009

Trustworthy Global Computing

View full text Add to dashboard Cite

show abstract

“…The references between different activities are in fact transitive references, since there can be a chain of local pointers between the active object and the remote reference. This is identical to the graph summarization technique in [4].…”

Section: The Reference Graphmentioning

confidence: 99%

“…In the DGC algorithm [4] by Veiga and Ferreira, cycle detection messages traverse the reference graph and grow information about it. Referencers are called dependencies and represent the still unknown part of the graph.…”

Section: Related Workmentioning

confidence: 99%

“…However, this DGC is unable to collect distributed cycles of garbage since it is based on a reference listing [3] approach. Recent cyclic distributed garbage collector [4] typically suffers from a large space complexity, because each process has a view of the whole distributed system. This paper presents a DGC geared towards the collection of distributed activities which is able to collect both acyclic and cyclic garbage, typically referred to as a complete DGC.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Garbage Collecting the Grid: A Complete DGC for Activities

Caromel

Chazarain

Henrio

2007

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Grids are becoming more and more dynamic, running parallel applications on large scale and heterogeneous resources. Explicitly stopping a whole distributed application is becoming increasingly difficult. In that context, there is a strong need to free resources a soon as they become useless, leading to automatic termination, using distributed garbage collecting techniques. We propose in this paper a new distributed garbage collector for active objects taking into account cycles but with a complexity similar to the distributed garbage collector of Java/RMI. The algorithm is based on a different approach to collect acyclic and cyclic garbage. On one hand, acyclic garbage is collected by knowing the immediate referencers of an active object and detecting the lack of these referencers. This behavior with respect to acyclic garbage is common to the distributed garbage collector of RMI. On the other hand, cyclic garbage is detected by considering the recursive closure of all the referencers of an active object and finding cycles of active objects waiting for requests. These cycles are found by letting idle active objects make a consensus on a common final activity. The algorithm is fully distributed and has been implemented with no modifications to the local garbage collector. Benchmarks have shown the scalability of the algorithm in a grid context.

show abstract

“…In this paper, the scheduling parameter is set to be 50 for all hybrid schemes, except for the schemes by [36], of which is dynamically adjustable according to cluster heterogeneity. First, execution time on the grid for GSS [29] group is investigated. Figure 3(a) illustrates execution time of static [28], dynamic [29] and our scheme, with input matrix size 512×512, 1024×1024 and 1536×1536 respectively.…”

Section: Application 1: Parallel Loop Self-scheduling For Matrix Multmentioning

confidence: 99%