Jose M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In PARLE'91---Parallel Architectures and Languages Europe, volume 505 of Lecture Notes in Computer Science, pages 150--165, Eindhoven (the Netherlands) , June 1991. Springer-Verlag.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....this introduces communication overhead and the GC algorithm is still not resilient to failures. For these reasons there are many variants of the reference counting algorithm that can be grouped in the following main categories: weighted reference counting [15, 123] indirect reference counting [58, 95], and reference listing [18, 85, 106, 108] Each one of these variants avoids the transmission of increment messages, therefore solving the two race problems previously described. We describe these variants now. 2.3.1.1 Weighted Reference Counting Each cross partition reference has two ....

Jos'e M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In PARLE'91---Parallel Architectures and Languages Europe, volume 505 of Lecture Notes in Computer Science, pages 150--165, Eindhoven (the Netherlands) , June 1991. Springer-Verlag.

....value that cannot be split. Several techniques have been proposed to overcome this problem. The trivial solution is to increase both the total weight at the owner and the partial weight at the client by the same amount. But this requires additional messages on the critical path. Other approaches [11, 3, 8] add some sort of hierarchy where the client gives out new weight without contacting the owner. But the new client has to return the weight to the issuing client instead of the owner. Hence, there is no additional message on the critical path but the collection process itself is nested across the ....

J. M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In Parallel Architectures and Languages Europe, number 365 in Lecture Notes in Computer Science, pages 150--165, Eindhoven, The Netherlands, June 1991. Springer-Verlag Berlin, Heidelberg, New York.

....world algorithms require costly termination mechanisms when facing distribution, whereas reference counting is completely defeated by common messages failures. In order to adapt those techniques to distributed environments, many recent proposals try to relax traditional invariants [Dickman 1992, Piquer 1991b, Watson 1987] whereas others rely on reliable communication protocols [Hughes 1985b, Lang 1992b, Mancini 1991] The former family algorithms is usually based on reference counting. Therefore they cannot garbage collect distributed cycles and must assume that such graphs are rare. The second ....

.... not resilient to message failures [Lang 1992b] may be completely defeated by space failures [Hughes 1985b] or fail to address large network [Liskov 1986b] Our protocol belongs to the former family and bears some similarities to a number of proposals based on reference counting [Dickman 1992, Piquer 1991b] Unlike those approaches, however, we maintain an entry item per source space that permits us to tolerate message loss whilst avoiding the dangers of duplicated messages. Dickman [Dickman 1992] proposes Optimizing Weighted References Counting improving traditional Weighted Reference Counting ....

[Article contains additional citation context not shown here]

Jos'e M. Piquer. Indirect referencecounting, a distributed garbage collection algorithm. In PARLE'91---Parallel Architectures and Languages Europe, volume I of Lecture Notes in Computer Science, pages 150--165, Eindhoven (the Netherlands) , June 1991. Springer-Verlag.

....collection is a difficult problem that has only been addressed partially. All known algorithms that can collect cycles are either not fault tolerant [Lang 1992a] can delay garbage reclamation [Hughes 1985a] or are not scalable [Liskov 1986a] On another hand, incomplete techniques [Dickman 1992, Piquer 1991a] are usually more resilient to messages failures, but these benefits rely on the assumption that distributed cycles are relatively rare. Thus, designing a distributed garbage collection poses a challenging problem: reclaiming all kinds of data structures while achieving fault tolerance, ....

....of data structures while achieving fault tolerance, scalability and efficiency. Our protocol (hereafter called SGP) Shapiro 1990] addresses this challenge but fails to collect cycles. It is based on a variant of reference counting and bears some similarities with a number of other proposals [Piquer 1991a, Dickman 1992] Author s other affiliation: Laboratoire d Informatique Th eorique et Programation (LITP) Universit e de Paris VI, 75252 Paris C edex 05 y This work is supported in part by the Minist ere de la Recherche et de la Technologie One of the requirements in designing this protocol ....

[Article contains additional citation context not shown here]

Jos'e M. Piquer. Indirect referencecounting, a distributed garbage collection algorithm. In Pro. 1991 Principles of Programming Languages, feb 1991.

....Counting is the limited number of duplications. In the worst case, an unnecessarily long chain of indirect references may be created. One solution is to extend the number of bits allocated to the weights, but this has a cost and does not completely avoid the creation of indirect references. Piquer [1991] suggests an original solution to this problem. The key idea of this algorithm is to encapsulate in an exit item two locators rather than one. A strong locator refers to an entry item in the sender space. An additional weak locator shortcuts the strong one, and refers ahead, to a better location ....

....Table 1: Taxonomy of some popular distributed GC techniques. main distributed GC characteristics Distributed GC cyclic floating large imple space message failure Techniques garbage scale mented failure duplic. loss late Reference Counting [Bevan, 1987] ffl [Dickman, 1991] ffi ffl ffl ffl [Piquer, 1991] ffl ffl ffl ffl [Goldberg, 1989] ffl Reference Listing [Shapiro et al. 1990] ffl ffl ffl ffl ffl ffl ffl [Birrell et al. 1993] ffl ffl ffl ffi ffl ffl ffl Tracing [Ali, 1984] ffl [Hughes, 1985] ffl ffl [Ladin and Liskov, 1992] ffl ffl [Lang et al. 1992] ffi ffl ffi ffl [Juul and Jul, ....

Jos'e M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In Eddy Odijik, M. Rem, and Jean-Claude Sayr, editors, Parallel Architectures and Languages Europe, number 365, 366 in Lecture Notes in Computer Science, pages 150--165, Eindhoven, the Netherlands, June 1991. Springer-Verlag.

....to evaluate the SGP protocol, we have prototyped it on a distributed Lisp, Transpive [12] implemented at INRIA, running on a multi Transputer board hosted by a Sun server machine. For the purpose of this evaluation, we replaced Piquer s original Indirect Reference Count (IRC) garbage collector [11], provided with Transpive, with a prototype implementation of the SGP protocol. SGP provides all the functionality of Piquer s GC, and in addition is resilient to message or site failures. The motivations for this approach are the following: ease of prototyping the algorithm in a functional ....

....task. Stop the world algorithms require costly termination mechanisms when facing distribution, whereas reference counting is completely defeated by common messages failures. In order to adapt those techniques to distributed environments, many recent proposals try to relax traditional invariants [2, 11, 15] whereas others rely on reliable communication protocols [3, 6, 10] The former family algorithms is usually based on reference counting. Therefore they cannot garbage collect distributed cycles and must assume that such graphs are rare. The second family ensure better liveness but all known ....

[Article contains additional citation context not shown here]

Piquer, J. M. Indirect reference-counting, a distributed garbage collection algorithm. In PARLE'91---Parallel Architectures and Languages Europe (Eindhoven (the Netherlands), June 1991), vol. I of Lecture Notes in Computer Science, Springer-Verlag, pp. 150--165.

.... in distributed computing environments too, with the emerging low latency interconnect technologies [4, 8] and global computing paradigms[1] We are now planning to: adapt our analysis to other GC algorithms such as Hughes time stamp based global collector[7] indirect reference counting[12], or generational local collector [11] perform more experiments using realistic applications and true distributedmemory computers, and extend our analysis of global mark and sweep. Acknowledgment A discussion with Takashi Chikayama was a start to extend our previous incomplete work. ....

Jose M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In Parallel Architectures and Languages Europe, Lecture Notes in Computer Science, number 365,366, pages 150--165. Springer-Verlag, June 1991.

....location cache. Some hotspots may arise, however, if particular well known objects are referenced from a great many remote spaces, due to the accumulation of scions. 4. 3 Measured Performance We have prototyped an earlier version of our protocol, called SGP [25] on the distributed Lisp Transpive [18]. This version lacks weak locators, uses a message protocol rather than call reply and uses an extra timestamp vector instead of timestamping stubs. A detailed account and analysis of this experiment may be found in [19] For our evaluation, we replaced Piquer s original distributed Indirect ....

....the face of failures. A common approach is to use reliable mechanisms to enforce strong consistency, which is expensive. A more recent approach is to relax traditional GC invariants. Our scheme is based on the latter alternative and bears similarities to some proposals based on reference counting [6, 18]. Unlike those approaches, however, a scion is maintained per source space, which permits us to tolerate message loss while avoiding the dangers of duplicated delete messages. Dickman [6] proposes an optimised weighted reference counting (oWRC) algorithm. In order to deal with unreliable ....

Jos'e M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In PARLE'91--- Parallel Architectures and Languages Europe, volume I of Lecture Notes in Computer Science, pages 150-- 165, Eindhoven (the Netherlands), June 1991. SpringerVerlag.

....counting and tracing. Most published algorithms combine counting and tracing to some degree. For instance in Dickman [7] most, but not all, unreachable scions are detected by counting; the remainder are detected by a complementary global trace. Counting variants are either non fault tolerant [1, 21, 29] or fault tolerant [26] see related Sections 3.1, 3.2) The former are of course simpler than the latter. They also differ on how they deal with cycles of garbage: by migration [4] by a complementary trace [7] or not at all [1, 21, 29, etc. Tracing is feasible only in a small scale system. ....

Jos'e M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In PARLE'91---Parallel Architectures and Languages Europe, volume 505 of Lecture Notes in Computer Science, pages 150--165, Eindhoven (the Netherlands), June 1991. Springer-Verlag.

....collected. The rest of this section analyzes the costs of our protocols and their tolerance to a the fault conditions that are common in actual distributed systems. Memory usage assumes stock 32 bit hardware. 4. 1 Failures The failure assumptions are weaker than in most comparable material [4, 5, 6, 15, 16, 21]. Messages may be duplicated, lost or delivered out of order. Processor pairs are subject to periods during which communication between them may be impossible; however, it is not assumed that this failure is either symmetric or transitive. By minimizing the assumptions we ensure our mechanisms are ....

....among processes of a single machine, the only case where the no crash assumption is reasonable. At this point, our cleanup protocol is reduced to reference counting. 4. 4 Measured Performance We have prototyped an earlier version of our protocol, called SGP [30] on the distributed Lisp Transpive [21]. This version lacks weak locators, uses a message protocol rather than call reply and uses an hts timestamp vector instead of timestamping stubs. A detailed account and analysis of this experiment may be found elsewhere [23] For our evaluation, we replaced Piquer s original distributed Indirect ....

[Article contains additional citation context not shown here]

Jos'e M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In PARLE'91---Parallel Architectures and Languages Europe, volume 505 of Lecture Notes in Computer Science, pages 150--165, Eindhoven (the Netherlands) , June 1991. Springer-Verlag.

....Counting is the limited number of duplications. In the worst case, an unnecessarily long chain of indirect references may be created. One solution is to extend the number of bits allocated to the weights, but this has a cost and does not completely avoid the creation of indirect references. Piquer [1991] suggests an original solution to this problem. The key idea of this algorithm is to encapsulate in a stub two locators rather than one. A strong locator refers to a scion in the sender space. An additional weak locator shortcuts the strong one, and refers ahead, to a better location of the ....

....Figure 17. Tracing within a group composed of spaces A and B. main distributed GC characteristics Distributed GC cyclic floating large imple space message failure Techniques garbage scale mented failure duplic. loss late Reference Counting [Bevan, 1987] ffl [Dickman, 1991] ffi ffl ffl ffl [Piquer, 1991] ffl ffl ffl ffl [Goldberg, 1989] ffl Reference Listing [Shapiro et al. 1990] ffl ffl ffl ffl ffl ffl ffl [Birrell et al. 1993] ffl ffl ffl ffi ffl ffl ffl Tracing [Ali, 1984] ffl [Hughes, 1985] ffl ffl [Ladin and Liskov, 1992] ffl ffl [Lang et al. 1992] ffi ffl ffi ffl [Juul and Jul, ....

José M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In Eddy Odijik, M. Rem, and Jean-Claude Sayr, editors, Parallel Architectures and Languages Europe, number 365, 366 in Lecture Notes in Computer Science, pages 150--165, Eindhoven, the Netherlands, June 1991. Springer-Verlag.

No context found.

Jose M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In PARLE'91---Parallel Architectures and Languages Europe, volume 505 of Lecture Notes in Computer Science, pages 150--165, Eindhoven (the Netherlands) , June 1991. Springer-Verlag.

No context found.

J. M. Piquer. Indirect reference-counting, a distributed garbage collection algorithm. In Parallel Architectures and Languages Europe, number 365 in Lecture Notes in Computer Science, pages 150--165, Eindhoven, The Netherlands, June 1991. Springer-Verlag Berlin, Heidelberg, New York.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC