V. V. Murty and V. K. Garg. Synchronous message passing. In Proceedings of the International Symposium on Autonomous Decentralized Systems, pages 208--214, Phoenix, Arizona, April 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... application level event is denoted send m:sender (m) ffl The deliver(m) primitive allows a process (say P j ) to receive a message that has been sent to it by an other process (so, P j 2 dest m ) The corresponding application level event is denoted deliver j (m) Previous proposals [14,12] consider a message is sent to a single destination process. Here, the send primitive allows to multicast a message to an arbitrary set of destination processes (this set is defined online by the sending process) 2.3. li Communication When a process executes send(m; dest m ) we say that it ....

....ffl Logical Instantaneity. Let IN be the set of natural integers. This set constitutes the (logical) time domain. Let H a be the set of all application level communication events of the computation. There exists a timestamping function TF from H a into IN such that 8(e; f) 2 H a Theta H a [14]: 2 Of course, for a message (that has been li sent) to be li delivered by a process P j 2 dest m , it is necessary that P j issues enough invocations of the deliver primitive. 5 (LI 1 ) e and f have been produced by the same process with e first ) TF (e) TF (f) LI 2 ) 8m : 8j 2 dest m : ....

Murty V.V. and Garg V.K., Synchronous Message Passing. Tech. Report TR ECEPDS -93-01, University of Texas at Austin, (1993).

....(m) ffl The deliver(m) primitive allows a process (say P j ) to receive a message that has been sent to it by an other process (so, P j 2 dest m ) The corresponding application level event is denoted deliver j (m) 3 Note that channels are not required to be fifo. 3 Previous proposals [16, 21] consider a message is sent to a single destination process. Here, the send primitive allows to multicast a message to an arbitrary set of destination processes (this set is defined online by the sending process) 2.3 li Communication When a process executes send(m; dest m ) we say that it ....

....m:sender. ffl Logical Instantaneity. Let IN be the set of natural integers. This set constitutes the (logical) time domain. Let H a be the set of all application level communication events of the computation. It exists a timestamping function T from H a into IN such that 8(e; f) 2 H a Theta H a [16]: LI 1 ) e and f have been produced by the same process with e first ) T (e) T (f) LI 2 ) 8m : 8j 2 dest m : e = send m:sender (m) f = deliver j (m) T (e) T (f) From the point of view of the communication of a message m, the event send m:sender (m) is the cause and the events deliver j ....

Murty V.V. and Garg V.K., Synchronous Message Passing. Tech. Report TR ECE-PDS-93-01, University of Texas at Austin, (1993).

....for Shared Memories 17 m 1 m 2 m 3 (a) Causally ordered communication m 3 m 1 m 2 (b) Not causally ordered communication P 1 P 2 P 3 Figure 5: Causal ordering m 1 m 2 m 3 m 1 m 2 m 3 P 1 P 2 P 3 Figure 6: Logically instantaneous communication PI n968 18 M. Raynal and A. Schiper [28], T is the timestamping function which assigns a timestamp T (x) to each event x) 9T : H 7 Nsuch that : a i b ) T (a) T (b) 8m : T (send(m) T (receive(m) 8.2.3 Relations Between Both Models Let b H be a history in the message passing model. 15] introduces the empty interval ....

V.V. Murty and V.K. Garg. Synchronous message passing. Technical Report ECEPDS -93-01, University of Texas at Austin, Dpt. of Elec. and Computer Engineering, 1993. PI n968 22 M. Raynal and A. Schiper

....pair of processes the receiving order is the same as the sending one. Causal order means that if two sends are causally related [9] and concern the same destination process,then messages are delivered in their sending order. Protocols implementing such communication modes have been proposed in [1, 14, 12, 11, 15]. In this paper we suppose a reliable asynchronous distributed system and we are interested in characterizing communications modes. Such characterizations are particularly useful whenone is interested in understanding distributed computations. We consider asynchronous distributed computations and ....

....such a way that, for each message, the send event and the receive one can be put on the same vertical line. More formally all events can be assigned timestamps such that time increases within each process and for each message its send and receive events are assigned the same and unique timestamp [11], i.e. 9T : E7 N such that ae aOE l b : T (a) T (b) aOE m b : T (a) T (b) The existence of such timestamps indicate there is a logically equivalent computation where communications are instantaneous. An example of such communications is shown if Figure 2. 3.2 Causally Ordered ....

[Article contains additional citation context not shown here]

V.V. Murty and V.K.Garg. Synchronous message passing. Technical Report ECE-PDS-93-01, University of Texas at Austin , Department of Electrical & Computer Engineering, October 1993.

....further. It is known that a synchronous computation, i.e. a computation based on synchronous messages, is logically equivalent to a computation in which all messages are instantaneous. In other words, we can always draw the time diagrams for synchronous computations with vertical message arrows [1, 16]. If we ignore internal events in a computation, then timestamping events is equivalent to timestamping messages in the computation. We define a partial order (M; 7 ) where M is the set of messages in the computation and 7 is the order relationship between messages (defined formally in Section ....

....Each process maintains a vector of size d, where d is the size of the edge decomposition. We assume that information about edge decomposition is known by all processes in the system. The online algorithm is presented in Figure 5. Due to the implementation of synchronous message ordering [16], we assume that for each message sent from P i to P j , there exists an acknowledgement sent from P j to P i . Essentially, to timestamp each message, the sender and the receiver must first exchange their vector clocks. Then, each process computes the component wise maximum between the local ....

V. V. Murty and V. K. Garg. Synchronous message passing. In Proceedings of the International Symposium on Autonomous Decentralized Systems, pages 208--214, Phoenix, Arizona, April 1995.

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