L.L. Peterson, N.C. Bucholz, and R.D. Schlichting, "Preserving and Using Context Information in Interprocess Communication," ACM Trans. Computer Systems, vol. 7, no. 3, pp. 217-246, 1989.  Home/Search   Document Details and Download   Summary   ACM   TOC   Related Articles   Check

....Asymmetric protocols ( 7, 9, 21] use a centralized coordinator for ordering the messages. The problems with this method are in the serial bottleneck they create at the coordinating site, and with the costly handling of faults in case the coordinator crashes. Existing symmetric protocols ([28, 6, 17, 24]) require all machines to relay their up todate view on the latest delivered messages in order for a new message to be admitted into the total order. These methods are essentially all ack, i.e. require all machines to send ack for a single message admittance. Unfortunately, all ack protocols ....

....messages m e;1 , m a;1 and m f;1 . If a message arrives at a machine, and some of its causal predecessors are missing, Transis transparently handles message recovery and re ordering. In this way, it guarantees that the ToTo protocol receives the messages in causal order. Other environments like [7, 28] are equally suitable for providing the causality requirement. Below, we sometimes refer to the environment and messages as the Transis environment and Transis messages. We think of the causal order as a directed acyclic graph (DAG) the nodes are the messages, the arcs connect two messages that ....

[Article contains additional citation context not shown here]

L. L. Peterson, N. C. Buchholz, and R. D. Schlichting. Preserving and using context information in interprocess communication. ACM Trans. Comput. Syst., 7(3):217--246, August 89.

....process, and occurs before , or (2) denotes the event when a process communicates some information and denotes the event when a process gets information , or (3) there exists an event , such that . Causal order has been proposed as a mechanism to build dependable systems [43, 8, 9, 39]. It ensures that if the proposal of an update happened before the proposal of an update is delivered by a process . Processes may communicate using services other than the group communication service, e.g. they could use a network file system. These other communication services ....

L. L. Peterson, N. C. Buchholz, and R. D. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217--246, Aug 1989.

....local area communication and hierarchical composition while in NAVTECH emphasis was given on dynamic configuration and filtering of local control data. Composition is an established principle of distributed protocol design. Typical examples of protocol composition frameworks are the # kernel[29], Horus [40] Ensemble [22] Coyote [6] The NAVTECH architecture is not tied to a concrete protocol composition framework, but it assumes that modules are implemented using such a construct. More recently, we are developing on the Appia framework[25] One of the early prototypes of NAVTECH used ....

....of NAVTECH protocols. Although the architecture itself is not tied to a concrete protocol composition framework, its implementation should desirably rely on a run time environment that allows micro protocols to be composed in a flexible way. Typical examples of such frameworks are the # kernel[29], Horus [40] Ensemble [22] Coyote [6] and Appia[25] The lessons learned from these implementations motivated the development of a new composition framework[30] This framework, called Appia[25] is currently being used to implement an instantiation of the NAVTECH architecture to support ....

L. Peterson, N. Buchholz, and R. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217--146, August 1989.

....used to implement a number of prototype applications. One example is the RNFS 26 (replicated NFS) file system, a network file system that is tolerant to fail stop failures and runs on top of NFS, that was designed using the state machine approach [Marzullo Schmuck 88] The Psync primitive [Peterson et al. 89] which has been implemented in the x kernel [Hutchinson Peterson 88] is similar to the CBCAST of ISIS. Psync, however, makes available to the programmer the graph of the message potential causality relation, while CBCAST does not. Psync is intended to be a low level protocol that can be ....

Peterson, L.L, N.C. Bucholz, and R.D. Schlichting. Preserving and using context information in interprocess communication. ACM TOCS 7, 3 (August 1988), 217-246.

.... Therefore, we have opted to base our new algorithm on the Hybrid Total Order algorithm described in [25] This algorithm combines two of the most used approaches to enforce total order in systems where a perfect failure detector is available, specifically: the token site [6, 17] and symmetric [24, 7] approach. It does so by implementing a dynamic configuration policy that adapts the algorithm behavior such that the most adequate mechanism is used as a function of the network delays and of the tra#c load. In the token based approach, a process is responsible for ordering messages on behalf of ....

....token holder) Thus, token based approaches are ine# cient in face of large network delays. In the symmetric approach, ordering is established by all processes in a decentralized way, using information about message stability. This approach usually relies on logical clocks [20] or vector clocks [4, 24, 19]: messages are delivered according to their partial order and concurrent messages are totally ordered using some deterministic algorithm. Symmetric algorithms have the potential for providing low latency in message delivery when all processes are producing messages. In fact, using a technique ....

L. Peterson, N. Buchholz, and R. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217--146, August 1989.

....latency. For instance, in a sequencer based protocol [5, 16] all processes (except the sequencer itself) have to wait for the message to reach the sequencer and for the sequence number to travel back before the message can be delivered. On the other hand, protocols based on causal history [18, 23, 10] can provide latency proportional to the interarrival delay of each sender and thus lower latency than sequencer based protocols. However, when each sender has a large in # Research supported by FCT, ESCADA proj (POSI 33792 CHS 2000) terarrival time and low latency is desired, this requires the ....

.... to receivers before suffering the latency of ordering, excluding algorithms which delay dissemination until messages are ordered [15, 3, 2] It is also required that the decided order is directly derived from the spontaneous ordering at some process, which is not true for causal history algorithms [18, 23, 10]. These requirements are satisfied by consensus based algorithms [7] as long as the coordinator for each instance of consensus is likely to be the same. 9 6. Conclusions Although total order multicast is a convenient tool in programming fault tolerant distributed systems, it exhibits higher ....

L. Peterson, N. Buchholz, and R. Schlichting. Preserving and using context information in interprocess communication. ACM Trans. Comput. Syst., 7(3), Aug. 1989.

....is typically performed using ordered reliable multicasts: reliability ensures that all correctly functioning members of a group receive messages intended for that group and ordering ensures that these messages are received in an identical order at each of the members. ISIS [4] X kernel Psync [5] and DELTA 4 [6] are examples of distributed systems relying on such multicasts for group management. In this respect, the group management scheme employed in Arjuna is different: we use unordered reliable multicasts at the communication level and impose ordering at the application level, by ....

L. Peterson, N.C. Buchholz and R.D. Schlichting, "Preserving and using context information in interocess communication", ACM TOCS, 7,3, pp. 217-246, Aug. 1989.

.... Therefore, we have opted to base our new algorithm on the Hybrid Total Order algorithm described in [25] This algorithm combines two of the most used approaches to enforce total order in systems where a perfect failure detector is available, speci cally: the token site [5, 17] and symmetric [24, 6] approach. It does so by implementing a dynamic con guration policy that adapts the algorithm behavior such that the most adequate mechanism is used as a function of the network delays and of the trac load. In the token based approach, a process is responsible for ordering messages on behalf of ....

....the token holder) Thus, token based approaches are inecient in face of large network delays. In the symmetric approach, ordering is established by all processes in a decentralized way, using information about message stability. This approach usually relies on logical clocks [20] or vector clocks [3, 24, 19]: messages are delivered according to their partial order and concurrent messages are totally ordered using some deterministic algorithm. Symmetric algorithms have the potential for providing low latency in message delivery when all 7 RTO1 Agreement: Consider RTO broadcast(m) If a correct ....

L. Peterson, N. Buchholz, and R. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217-146, August 1989.

....GBCAST is executed in the same state everywhere requires that all nodes in the network participate in the synchronization part of the protocol. In contrast, our method uses fewer messages and requires that only the service nodes participate in the global synchronization. The Psynch protocol [29] is an IPC mechanism that supports causal message ordering by explicitly encoding this ordering in each message. It operates in the presence of network and process crashes and can be viewed as an optimized implementation of ISIS CBCAST that wansmits message iris instead of messages and thus cuts ....

Peterson, L. L., Bucholz N. C., and Schlichting, R.D.. Preserving and using context information in interprocess communication. ACM Trans. on Computer Systems 7(3):217-246, August, 1989.

....de live r ( interface primitive defined by the layers above. 3. 1 Group Communication Protocol The Group Communication Protocol required by the design of the Replica Synchronisation Protocol resembles the properties of typical group cast or reliable multicast protocols from the literature [4,5,12,13,14,31]. In particular, existing systems such as Horus [4] Transis [5] or RELACS [31] could be used to provide the functionality of GCP. The exact properties of the GCP layer are outlined in the following paragraphs. By the terms multicast and deliver , we refer respectively to ....

L. Peterson, N. Bucholz, and R. Schlichting, "Preserving and Using Context Information in Inter-process Communication," ACM Transactions on Computer Systems, vol. 7, 1989.

....mechanisms provide an abstraction through which such processes exchange messages. Various mechanisms for one to one communication, e.g. datagrams, virtual circuits, remote proce dure calls [4, 10] F channels [1] have been studied. Mechanisms for broadcast communication have also been studied [2, 3, 7, 11, 14]. These mechanisms provide primitives which support abstractions helpful in designing distributed algorithms. F channels have been proposed as a mechanism for one to one communication between pairs of processes in a distributed system [1] An F channel is a single non FIFO channel that ensures ....

....The overhead, both in terms of the amount of information carried by a messages and the amount of information kept at each site, is independent of the number of processes in the system. It must be noted that implementations of causal broadcast in related works, e.g. ISIS [2, 3, 11] and Psync [14] have un bounded overheads in terms of information carried by each message. This is due to the unbounded clocks used to encode this information. The counter based implementation of the GS primitive suggested in [5] also has unbounded overhead in terms of both, the information carried by each ....

[Article contains additional citation context not shown here]

L.L. Peterson, N.C. Buchholz, and Schliching R.D. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217-246, August 1989. 19

.... systems motivated this work, and are described in [2, 8] Atomic broadcast has been traditionally used to achieve fault tolerance by replicating functionality (e.g. using the state machine approach [18] and previous work on atomic broadcast algorithms are in the context of fault tolerance [10, 6, 11, 16, 15, 14, 5, 9, 1]. In these works, one main concern is to guarantee total order and agreement when the broadcaster fails during the broadcast. To obtain this guarantee, a recipient needs to send messages to other recipients, either to inform them of a previously delivered message, or to inform them of the intent ....

L. L. Peterson, N. C. Buchholz, and R. D. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217--246, Aug. 1989.

....is a range of applications, in the real time, input output area, where protocols featuring very small and oe are required, to support lock step, time triggered type of operation [13] Clearly, clock driven protocols are the solution. Alternatively, in application independent distributed systems [3,11,16], running non real time applications, clock less protocols are definitely very well suited, due to the performance advantage. However, clock less protocols also have their place in real time applications, when the timeliness assurances in communication and processing required, match their ....

Larry L. Peterson, Nick C. Buchholdz, and Richard D. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3), August 1989. 6

....r. Causal delivery ensures that if send p (m 1 ) Gamma p send q (m 2 ) then deliver r (m 1 ) Gamma p deliver r (m 2 ) i.e. m 1 is delivered to r before m 2 . 2 Well known causal delivery protocols are the causal broadcast multicast protocols using: piggybacking [2] context graphs [12], or vector clocks [3, 14] These protocols, adequate for asynchronous systems, deliver messages according to a logical ordering [2] a message m 1 is said to logically precede ( l ) m 2 if m 1 is sent before m 2 , by the same participant or m 1 is delivered to the sender of m 2 before it ....

Larry L. Peterson, Nick C. Buchholz, and Richard D. Schlichting. Preserving and Using Context Information in Interprocess Communication. ACM Transactions on Computer Systems, 7(3), August 1989.

....developed using layers of increasing complexity. Causally ordered multicasts often are considered as underlying level protocols. Then, a total order is built using the service offered by the causal protocol to provide a total and causal order. This approach is used for both broadcast protocols [6] [7] and multicast protocols [1] It needs to exchange in each message a set of historical data which can be large even if some optimizations are used. Moreover, the totally ordered layer does not take benefit from the presence of the causal layer. Our formal result gives an easier solution to ....

....works This protocol is similar to [3] and [5] In [5] the server of order moves from one process to another one by a token passing. But for the authors [3] and [5] the given protocol provides only a total order (T) Here we prove it is a causal (C) and total (T) broadcast. The algorithms of [7] and [6] are close and need two layers of protocol to build a causal and total broadcast. The total order (T) consists of a topological sort. In contrast, we need only a total layer to achieve a causal and total broadcast. 4.2 Causal and total multicast The multicast protocol uses, as in the ....

L. L. Peterson, N. C. Bucholz, R. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):213-246, August 1989.

.... protocols algorithms on reliable multicast are surveyed in the next two sections, such as Isis and Horus systems of Birman [1, 2] token list [3] and its implementation [4] sequencer [5, 6] two phase [7, 8] tree [9] logical token ring [1, 10, 11] 3 phase [12] discrete acknowledgments [13, 14] and module [2, 15, 16] etc. We classify the well known algorithms as token passing, discrete acknowledgement and two phase approaches and briefly describe their features and associated problems. 1.1. Token passing approach Among the existing multicast techniques, the token passing approach is ....

....to the token passing algorithms, for a specific multicast message, a member determines the message (delivery) order and its atomicity only after gathering enough acknowledgements with the subsequent multicast messages. The subsequent multicast messages form a partial order (context graph in Psync [14] and ACK, NACK in Transis [20] carrying the positive or negative acks about the previously received messages. Based on the partial order (context graph) from ack, a site sees a relationship among the messages and therefore is able to determine a total order among the multicast messages. This ....

Peterson, L. L., Buchholz, N. and Schlichting, R. (1989) Preserving and using context information in interprocess communication. ACM Trans. Computer Syst., 7, 217--246.

....based on replication. For example reliable multicast with causal ordering is the basic operation that is used in the state machine approach to building fault tolerant distributed systems [46] Reliable multicast is provided by various communication systems, including Isis, Horus, Psync and Transis [6, 7, 40, 4]. Transactions have been adopted as a basic mechanism in some distributed programming languages for supporting the building of fault tolerant applications. This was pioneered in the Argus distributed language [33] and subsequently in object oriented distributed languages such as Avalon C [18] ....

L. L. Peterson, N. C. Bucholz, and R. D. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217--246, 1989.

....open loop mechanisms. 1.4.2 Multipeer Another style of interaction is multipeer, conveying the notion of spontaneous, symmetric interchange of information, amongst a collection of peer entities. This paradigm appeared as early as in [75] where it is called multipoint association, and also in [71] where it is called conversation, a term that we avoid in order not to cause confusion with a di erent paradigm with the same name described in [19] and also discussed below. Multipeer interactions are the kind of interaction one might wish among managers of a distributed database, a group of ....

....distributed applications. For instance, total delivery order is a requirement for the implementation of replicated statemachines [86] which is a general paradigm for implementing fault tolerant distributed applications. Although several protocols have been described in the literature [4, 16, 13, 26, 36, 50, 51, 55, 62, 63, 71], few were speci cally targeted to operate in (geographically) large scale systems. In a large scale network processes trac patterns are usually heterogeneous. The same applies to the network links: some processes will be located within the same local area network whereas others will be connected ....

[Article contains additional citation context not shown here]

L. L. Peterson, N. C. Buchholz, and R. D. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217-246, 1989.

....of any two messages such that one could have caused the other, that is, messages that are potentially causally related. The most well known logical precedence implementations are the causal broadcast multicast type of protocols using: logical clocks [15] piggybacking [1] context graphs [18], or vector clocks [2] 2 We distinguish received, the act of a message arriving at a site, from delivered, the act of it being given to the participant, since those instants may not be equivalent, for some ordering protocols. 3 Or in the universe, for that matter. 7 These protocols are non ....

Larry L. Peterson, Nick C. Buchholdz, and Richard D. Schlichting. Preserving and Using Context Information in Interprocess Communication. ACM Transactions on Computer Systems, 7(3), August 1989.

....sensor. Atomic Broadcast allows processes to reliably broadcast messages, so that they agree on the set of messages they deliver and the order of message deliveries. Applications based on these paradigms include SIFT [WLG 78] State Machines [Lam78,Sch90] Isis [BJ87, BCJ 90] Psync [PBS89] Amoeba [Mul87] Delta 4 [Pow91] Transis [ADKM91] HAS [Cri87] FAA [CDD90] and Atomic Commitment. Given their wide applicability, Consensus and Atomic Broadcast have been extensively studied by both theoretical and experimental researchers for over a decade. In this thesis, we focus on ....

L. L. Peterson, N. C. Bucholz, and Richard D. Schlichting. Preserving and using context information in interprocess communication. In ACM Transactions on computer systems 7,3, pages 217--246, August 1989.

....sensor. Atomic Broadcast allows processes to reliably broadcast messages, so that they agree on the set of messages they deliver and the order of message deliveries. Applications based on these paradigms include SIFT [WLG 78] State Machines [Lam78, Sch90] Isis [BJ87, BCJ 90] Psync [PBS89] Amoeba [Mul87] Delta 4 [Pow91] Transis [ADKM91] HAS [Cri87] FAA [CDD90] and Atomic Commitment. Given their wide applicability, Consensus and Atomic Broadcast have been extensively studied by both theoretical and experimental researchers for over a decade. In this paper, we focus on ....

Larry L. Peterson, Nick C. Bucholz, and Richard D. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217--246, August 1989.

No context found.

L. L. Peterson, N. C. Buchholz and R. D. Schlichting, `Preserving and using context information in interprocess communication', ACM Trans. Computer Systems, 7, 217--246 (1989).

No context found.

L.L. Peterson, N.C. Bucholz, and R.D. Schlichting, "Preserving and Using Context Information in Interprocess Communication," ACM Trans. Computer Systems, vol. 7, no. 3, pp. 217-246, 1989.

No context found.

Larry L. Peterson, Nick C. Bucholz, and Richard D. Schlichting. Preserving and using context information in interprocess communication. ACM Transactions on Computer Systems, 7(3):217--246, August 1989.

No context found.

PETERSON, L. L., BUCHHOLZ, N. C., AND SCHLICHTING, R. D. 1989. Preserving and using context information in interprocess communication. ACM Trans. Comput. Syst. 7, 3, 217--246.

First 50 documents  Next 50

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