K. Birman, R. Cooper, T. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The ISIS System Manual, Version (2.1), September 1990.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....computing. 2.2.1 Characterization of Message Passing Primitives for PDC In oder to identify the HCP services for PDC, we first study the primitives provided by some current parallel distributed programming tools. The software tools studied include EXPRESS [5] PICL [3] PVM [6] ISIS [1], the iPSC communication library [4] and the CM5 communication library (CMMD) 7] These tools were selected because of their availability at the Northeast Parallel Architecture Center at Syracuse University and also the following two reasons: 1) they support most poten tial computing ....

K. Birman, R. Cooper, T. Joseph, K. Kane, and F. Schmuck. The ISIS System Manual.

....are not guaranteed to arrive in the same order at all sites. On the other hand, MMConf is a toolkit, whereas the groupware features of our system are currently integrated in the application. DistEdit [Knis90] is a toolkit for programming multi user text editors. It relies on the ISIS toolkit [Birm89] for the distributed aspect of the architecture. This may cause performance problems, as stated by the authors, because of the concurrency control algorithms implemented by ISIS. GroupKit [Rose92] also is a toolkit. It runs on top of Interviews [Lint89] and provides groupware specific user ....

. Birman, K., Cooper, R., Joseph, T., Kane, and K., Schmuck, F., The ISIS System Manual, June 1989.

.... most of their solutions guarantee neither data consistency nor fault tolerance [11] Motivated by this gap between theory and practice, recent proposals for replicated databases [1, 2, 21, 18, 12, 15, 14, 17] propose new approaches that exploit the rich semantics of group communication systems [10, 19] to implement an eager style replica control. Most of these solutions propagate the updates of the transactions using a total order multicast that delivers all messages at all sites in the same order. The database uses this order as a pattern to follow in the case of conflicts, i.e. conflicting ....

....an asynchronous system where neither message delays nor computing speeds can be bounded with certainty. Messages may be lost and sites may fail by crashing (we exclude Byzantine failures) Crashed sites may recover. Sites are equipped with a group communication system supporting virtual synchrony [10, 19]. Virtual synchrony provides applications with the notion of group membership and with a reliable multicast communication primitive (a message is sent to all members of the group) Virtual synchrony provides consistent information about the set of group members that appear to be currently ....

[Article contains additional citation context not shown here]

K. Birman, R. Cooper, T. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The ISIS - system manual, Version 2.1. Technical report, Dept. of Computer Science, Cornell University, Sept. 1993.

.... none communication and in the same order. Symmetrically, replicated) messages sent by the replicas themselves should be filtered so that only a single copy of each replicated message is actually issued to the rest of the system this is known as n to 1 communication[11] 3 Isis Isis[3], a toolkit that goes a long way in the implementation of the active replication just described, was the original inspiration for Drago. In Isis, programmers can define groups of processes and then refer to them by a single name. Communication with a group of processes is by means of (different ....

Birman, K., R. Cooper, T. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The Isis System Manual. Version 2.1. September 1990.

....massage passing, failure detection, consensus, and so support building reliable distributed applications. Several solutions have already been proposed for structuring dependable distributed applications, but none has fully succeeded in bringing power and simplicity, together with adaptability ([5], 10] 13] 14] Only one of them, BAST ( 9] does provide ready to use components that, on the one side, hide the complexity of underlying distributed protocols, and on the other, allow to build fault tolerant distributed software. Although BAST provides several abstractions helpful in ....

....of it in distributed environment is yet a very complex or sometimes even impossible task. This is not surprising, as the implementation of the group abstraction often requires dealing with many non trivial issues, among which are: reliable communication, failure detection, consensus, etc. 1] [5], 7] 16] In 1985 it was shown by Fisher, Lynch and Paterson that the Consensus problem 1 has no deterministic solution in asynchronous distributed systems that are subject to even a single process crash failure. This assertion is widely known as the Fisher Lynch Paterson (FLP) ....

Birman K., Cooper R., Joseph T., Marzullo K., Makpangou M., Kane K., Schmuck F., Wood M., The Isis System Manual, Dept. Of Computer Science, Cornell University, Sept 1990

....as described in [1] including the environment assumed, will be brie y presented, and then we will introduce the re nements which permit the derivation of the more ecient protocol. A common way to build fault tolerant distributed application nowadays is by means of a reliable broadcast mechanism [9]. In fact, various fault tolerant broadcast protocols have been designed for their use in distributed systems. Most of these protocols insure order and reliable delivery of messages [3] 17] and some of them also preserve causality [2] On the other hand, order, reliability, and causality are ....

Birman K., Cooper R., Joseph T., Marzullo K., Makpangou M., Kane K., Schmuck F. and Wood M. September, 1990. The ISIS System Manual, Version 2.1.

....of fault tolerant distributed applications based on the active replication paradigm [18] The library, called Group IO [10] offers a simple interface to the implementation of reliable, atomic, causal, and uniform multicast. The work on Group IO has been motivated by our experience with Isis [3] and similar reliable multicast frameworks. The library allows also client server interactions where the client may be a group this interaction is not supported by ISIS and relies on an own consensus protocol [8, 9] to implement the uniform broadcast protocols. Group IO is the base on ....

....times. offered by Group IO, followed by another section with some programming examples. Three more sections then discuss implementation aspects of Group IO, and its relation with Ada 95 and Drago. The paper closes with some conclusions, and with references to related work. 2 ISIS ISIS [3] is a toolkit that goes a long way in the active replication line just described and which has been the original inspiration for Group IO. In ISIS programmers can define groups of processes and then refer to them by a single name. Communication with a group of processes is by means of (different ....

[Article contains additional citation context not shown here]

Birman, K., R. Cooper, T. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The Isis System Manual. Version 2.1. September 1990.

....of running applications, thus enabling cooperative computing over a network [22] The performance of the current prototype is comparable to that of Sun RPC. Parsons [27] has compared the usability and performance of Concert C with other tools for distributed programming including Isis [10] and PVM [32] We are adding full interoperability with OSF DCE, are currently designing Concert C , and are investigating preliminary designs for Concert Fortran. We are interested in developing tools for process management and control, such as a distributed visualizer debugger, and utilities ....

Kenneth P. Birman, Robert Cooper, et al. The ISIS system manual, version 2.0. Technical report, CS Department, Cornell, March 1990.

....six hexadecimal numbers, one per byte, separated by colons (for instance, 0c:00:07:21:00:85) See [23] for a description of the Ethernet addressing mechanism. Isis, in turn, uses character strings for group addresses (like, for instance, Group 1 ) addressing mechanisms of Isis are described in [2]. To override these difficulties, Lower Layer provides two views for addresses see code fragment 1: ffl An abstract tagged type, called Address, which will be extended by inheritance. Each LLM will provide an actual type for specifying addresses of the corresponding Com Protocol, including ....

K.P. Birman, R. Cooper, T.A. Joseph, K.P. Kane, and F.B. Schmuck. The Isis System Manual, version 2.1. Cornell University, September 1990.

....recomputes the state representation for its autonomous system, and advertises it. This is done regardless of how recently it received updates from other systems. We will investigate the use of a multicast facility for this channel, such as described by Birman, et al. in the isis system manual [3], and by Deering in [10] Provision is also made for polled updates, whereby a system can query another as to its status through the control channel and receive a reply through the update channel. Update cycles cannot be allowed in the communications structure of a system. An update cycle occurs ....

K. Birman, R. Cooper, T. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The ISIS System Manual, Version (2.1), September 1990.

....tools. These software tools are a subset of existing tools and our study is by no means comprehensive; however, it serves our purpose in the sense that it provides us with a list of services that are required to support HPDC. The software tools studied include EXPRESS [5] PICL [3] PVM [6] ISIS [2], the iPSC communication library [4] and the CM5 communication library (CMMD) 7] These tools were selected because of their availability at the Northeast Parallel Architecture Center at Syracuse University and also the following two reasons: 1) they support most potential computing ....

.... extest exhandle exchange exvsend exvreceive PICL send0 recv0 message0 recvinfo0 probe0 PVM ISIS iPSC Library CM5 Library CMMD CMMD send CMMD receive CMMD msg pending CMMD swap CMMD send and receive CMMD send v CMMD receive v snd rcv vsnd vrcv rcvinfo rcvmulti probe probemulti [x]bcast [2] [x]bcast l [2] reply reply l Group Communication exbroadcast excombine exconcat bcast0 g[op]0 [1] csend crecv isend irecv hsend hrecv csendrecv isendrecv g[type] op] 3] gopf gcol CMMD distrib to nodes CMMD receive bc from host CMMD reduce type CMMD scan type CMMD concat with nodes ....

[Article contains additional citation context not shown here]

K. Birman, R. Cooper, T. Joseph, K. Kane, and F. Schmuck. The ISIS System Manual.

....tools. These software tools are a subset of existing tools and our study is by no means comprehensive; however, it serves our purpose in the sense that it provides us with a list of services that are required to support HPDC. The software tools studied include EXPRESS [9] PICL [4] PVM [3] ISIS [10], the iPSC communication library [11] and the CM5 communication library (CMMD) 12] These tools were selected because of their availability at the Northeast Parallel Architecture Center and Syracuse University and for the following two reasons: 1) They cover most potential computing ....

K. Birman, R. Cooper, T. Joseph, K. Kane, and F. Schmuck, The ISIS System Manual.

....here in as far as local checkpointing is an essential part of distributed checkpointing and also of certain styles of distributed computation. Virtual time for example relies heavily on local checkpoints for distributed synchronisation. Many application frameworks require a checkpointing service[1,6,7,8,9,10,11,12,13,14,15,18]. Its implementation however, without direct OS support, can incur substantial costs in terms of development time and runtime overhead. On the other hand, OS kernels, be they micro or monolithic, implement process management functions such as scheduling and context switching as efficiently as ....

.... simulation[18,19] 1 Matching Operating Systems to Application Needs Wanted: An Application Aware Checkpointing Service 2 W2 94 Motivations for checkpointing include: concurrency control and atomicity[1,6,12] rollback and backtracking[2,6,12,15,18] failure recovery and fault tolerance[6,11,12,20], process and object migration[21,22] reversible execution[2,17] 3. State Management in a Virtual Time Framework Virtual time frameworks are of particular interest because they make the heaviest demands on state management. We distinguish between implicit virtual time (IVT) for example, ....

BIRMAN, K., COOPER, R., JOSEPH, T., MARZULLO, K., MAKPANGON, M., KANE, K., SCHMUCK, F., & WOOD, M., The ISIS System Manual, Version 2.1. , The Isis Project (1990).

....continuity and smoothness. Many systems enforce full temporal order of asynchronous events with tight synchronization but without continuity guarantees. Examples of these include schedulers used in distributed event simulations [5, 48, 73] and causality preserving protocols such as Birman s ISIS [8, 9]. Our research problem can be formulated as enforcing playback continuity over an asynchronous event simulation. Several systems enforce full temporal order of asynchronous events with synchronization and continuity but over coarse event grain. These include schedulers in multimedia authoring ....

K. Birman et al. The ISIS System Manual, Version 2.0, April 1990.

....computing. 2.2.1 Characterization of Message Passing Primitives for PDC In oder to identify the HCP services for PDC, we first study the primitives provided by some current parallel distributed programming tools. The software tools studied include EXPRESS [5] PICL [3] PVM [6] ISIS [1], the iPSC communication library [4] and the CM5 communication library (CMMD) 7] These tools were selected because of their availability at the Northeast Parallel Architecture Center at Syracuse University and also the following two reasons: 1) they support most potential computing ....

K. Birman, R. Cooper, T. Joseph, K. Kane, and F. Schmuck. The ISIS System Manual.

....broadcast in distributed systems is a topic that has been studied extensively for more than a decade [12] In fact, there are a number of existing projects and systems that provide a reliable transport layer as well as other services for distributed computing. Examples are the V system [6] ISIS [7], Psync [16] Amoeba [19] Trans [13] Transis [1] and Totem [2] However, we have observed that the properties required from the usercommunication layer associated with reliable broadcast protocols for distributed systems are different from the properties of the user communication layer ....

K. Birman, R. Cooper, T. A. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck and M. Wood, The ISIS System Manual, Dept. of Computer Science, Cornell University, September 1990.

.... 2 1 Introduction Interprocess communication by message passing is a crucial point in distributed systems and has been addressed in several RD13 notes [1] 2] Since UNIX does not provide all the required features, dedicated packages have to be used. Up to now, ISIS [3] was used in the RD13 DAQ software. The evolution of the package towards more and more sophisticated features and its monolithic architecture leading to too large binaries for front end DAQ, added to the fact that it is a licensed product pushed us to evaluate other public packages. The Message ....

K.P. Birman et al., THE ISIS SYSTEM MANUAL, Version 2.1

.... fast as the software implementation of broadcast on top of ATM that is presented in [14] For example, a broadcast of a 4Kbyte message on 8 machines takes about 6 msecs in our implementation compared to 15 msecs in the implementation in [14] We note there have been many existing systems, such as [4, 20, 1], that provide reliable transport protocol and other services for distributed computing. Our URTP protocol distinguishes itself from previous ones because it is targeted for supporting parallel computing using MPI programs and can take advantage of the global program semantics derived from our ....

K. Birman, R. Cooper, T. A. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck and M. Wood, The ISIS System Manual, Dept. of Computer Science, Cornell University, September 1990.

.... Examples of these systems include Arjuna [99] 100] a system for supporting the low level distribution needs of applications; Rendezvous [77] which provides support for sharing of views and data structures; DistEdit [57] a domain specific tool for constructing shared editors based on the ISIS [4] framework; and SUITE [18] 20] a system that provides support for several dimensions of coupling between replicated data structures and views, based on the SUITE user interface management system [21] 22] Very few collaboration support environments, however, have addressed the information ....

Birman, K., Cooper, R., Joseph, T., Kane, K., Schmuck, F. The ISIS System Manual, June 19, 1989.

.... favorably with other systems for parallel programming, like pvm [33] and C linda [15] as studied in [13] The measurement program is in the Concert C Tutorial [17] Parsons [26] has compared the usability and performance of Concert C with other tools for distributed programming including Isis [9] and PVM [33] 7 Future Work Concert C is the first of a set of compatible extensions to important programming languages. Thus, we may build Concert Cobol, Concert Fortran, Concert C , etc. Each of these languages will implement a distributed computing model called the process model: a ....

....the internal function signature representation as an intermediate form to mediate the search for equivalent representations in different languages. We are also investigating improvements to the run time to add group communication and fault tolerant primitives such as causal and atomic multicast [9]. The Unix implementations of Concert C, together with comprehensive documentation, including a tutorial [17] a programmer s manual [3] and example code, are available via anonymous ftp 8 . 7 We measured this performance between IBM RS 6000 workstations connected by a 16Mbit token ring, and ....

Kenneth P. Birman, Robert Cooper, et al. The ISIS system manual, version 2.0. Technical report, CS Department, Cornell, March 1990.

....made recoverable; 3) at the end of a distributed transaction a two phase commit is initiated and at least one record must be force written, whereas OR continuously logs information to stable storage and continuously commits computations whose dependencies are all logged. 6. 2 ISIS ISIS [12] is a distributed programming environment that has fault tolerance as one of its goals. It provides the user with a variety of tools, including ones that allow the replication of a user defined part of the process state, the maintenance of process groups, and so on. Using these tools, a ....

Kenneth P. Birman, R. C., et al. The ISIS system manual, version 2.0. Tech. rep., CS Department, Cornell, March 1990.

No context found.

K. Birman, R. Cooper, T. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The ISIS System Manual, Version (2.1), September 1990.

No context found.

K.P. Birman, R. Cooper, T.A. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The Isis System Manual. Dept. of Computer Science, Cornell University, September 1990.

No context found.

K.P. Birman, R. Cooper, T.A. Joseph, K. Marzullo, M. Makpangou, K. Kane, F. Schmuck, and M. Wood. The Isis System Manual. Dept. of Computer Science, Cornell University, September 1990.

No context found.

K. Birman, R.Cooper, T.A.Joseph, K.Marzullo, M.Makpangou, K.Kane, F.Schmuck, and M.Wood. ISIS Systems Manual, Version 2.0. Cornell University.

First 50 documents

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