D. Johansen, R. V. Renesse, and F. B. Schneider. An introduction to the tacoma distributed system version 1.0. Technical Report 95-23, University of TromsO, Norway, June 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....JoCaml constructs, their syntax, typing, and informal semantics, rst in a concurrent but local setting (Section 2) and nally in a distributed setting (Section 3) Starting from Objective Caml. High level distributed programming mostly relies on scripting languages (Agent Tcl [11] TACOMA [14], Telescript [29] Such languages are often specialized for some speci c task or architecture, and may o er poor performances for other operations, typically relegated to external calls in other languages. Besides, for the sake of exibility, these languages don t provide much structure, such as ....

D. Johansen, R. V. Renesse, and F. B. Schneider. An introduction to the tacoma distributed system version 1.0. Technical Report 95-23, University of TromsO, Norway, June 1995.

....Alice in this situation is to appeal to a higher authority in this case Bob someone whose policies are ranked higher than Charlie s. Bob may write a policy to evice Alice s agent, overriding Charlie s wishes. 4 Related Work There have been many proposed mobile agent systems, e.g. TACOMA [12] (Troms And COrnell Moving Agents) Agent TCL [6] and Telescript [16] Similarities with our work include: mobile agents on mobile devices (PDAs [9] and mobile phones [10] in TACOMA and the concept of regions (similar to our entities) in Telescript. Unlike our work, none of these previous systems ....

Dag Johansen, Robbert van Renesse, and Fred B Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical report, Department of Computer Science, University of Troms, Norway, 6 1995.

....We argue that these different concepts and notions cannot be compared directly. It is as inappropriate and misleading as trying to compare the emacs editor, the fork Unix system call, and the clientserver design paradigm. There is also confusion about terminology. For instance, several systems [17], 18] claim to be able to move the state of a component along with its code. This assertion is justified by the availability of mechanisms that allow the programmer to pack some portion of the data space of an executing component before the component s code is sent to a remote destination. ....

....a software component that is able to move between different execution environments. This definition has actually different interpretations. For example, while in Telescript [19] an agent is represented by a thread that can migrate among different nodes carrying its execution state, in TACOMA [17] agents are just code fragments associated with initialization data that can be shipped to a remote host. They do not have the ability to migrate once they have started their execution. On the other hand, in the artificial intelligence community the term agent denotes a software component that ....

[Article contains additional citation context not shown here]

# D. Johansen, R. van Renesse, and F.B. Schneider, "An Introduction to the TACOMA Distributed System---Version 1.0," Technical Report 95-23, Dept. of Computer Science, Univ. of Troms and Cornell Univ., Troms, Norway, June 1995.

....ends that led to the aforementioned attacks. Additional protocols e.g. published in [11, 15] remain to be investigated with regard to the applicability of the attacks described in [1] The representation of a mobile agent as a collection of files matches the Briefcase abstraction in TACOMA [16], and shares the same advantages. Most notably, the representation is independent of a particular agent implementation language. Our contribution is the addition of a security layer that is tailored to the needs of mobile agents. Performance 20 40 60 80 100 120 140 0.0 0.5 1.0 1.5 2.0 2.5 ....

D. Johansen, R. van Renesse, , and F. B. Schneider, "An introduction to the TACOMA distributed system version 1.0," Technical Report 95-23, Department of Computer Science, University of Troms, June 1995.

....than message or RPC based techniques, emphasizing the potential for mobile agents to be a uniform paradigm for developing distributed applications. Agents also remove the need for distributed applications to have their own control language, and multiple language systems such as D Agents, Tacoma [JvRS95] and Ara [PS97] demonstrate that the mobile agent paradigm is independent of language choice. D Agents is an excellent prototyping tool for distributed applications, but several challenges must be overcome to fully realize the six mobile agent benefits. Specifically, the protocol overhead of the ....

D. Johansen, R. van Renesse, and F.B. Schneider. An introduction to the TACOMA distributed system version 1.0. Technical Report 95-23, University of Troms, June 1995.

....applications, like mobile agents. These are different concepts and notions that cannot be compared directly. It is almost like comparing the emacs editor with the fork UNIX system call and the client server design paradigm. There is also confusion on terms. For instance, several systems [54, 83] claim to be able to move the state of a component together with its code, based on the fact that they allow the programmer to pack some portion of the data space of an executing component before the component s code is sent to a remote destination. This is quite different from the situation where ....

....a software component that is able to move between different execution environments. This definition has actually different interpretations. For example, while in Telescript [122] an agent is represented by a thread that can migrate among different nodes carrying its execution state, in TACOMA [54] agents are just code fragments associated with initialization data that can be shipped to a remote host, without the ability to migrate again while in execution. On the other hand, in the artificial intelligence community the term agent denotes a software component that is able to achieve a ....

[Article contains additional citation context not shown here]

Johansen, D., van Renesse, R., and Schneider, F. B. An Introduction to the TACOMA Distributed System---Version 1.0. Tech. Rep. 95-23, Dept. of Computer Science, Univ. of Troms and Cornell Univ., Troms, Norway, June 1995.

....can be dynamicly exchanged to match the desired behaviour. Mobile structure knowledge is also used in the context of multimedia databases [15] in Active Networks, and in the JINI architecture. A large number of projects deals with mobile code and mobile agents. Mobile agent systems like TACOMA [6], and ffMain [8] show different ways to build mobile code and mobile agent systems. Some of the ideas given there are used in INDIGO. 6 Conclusion and Future Work In this paper INDIGO was presented, an approach to architectures for digital libraries. It allows for the seamless integration of new ....

Dag Johansen, Robbert van Renesse, and Fred B. Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical Report 95--23, June 1995.

....the transmitted code, and the initiator of the transmission. Knabe [39] identifies four different kinds of transmissible code representations source code, interpreted intermediate representation, compiled intermediate representation, and machine code. REV [65] Avalon Common Lisp [20] TACOMA [36], and SafeTCL [14] transmit source code, which is executed via interpretation by the receiving node. Obliq [16] Telescript [73] and Java [25] transmit interpreted intermediate representations, and Omniware utilizes a compiled intermediate representation. Finally, Emerald CHAPTER 7. RELATED WORK ....

Johansen, D., van Renesse, R., Schneider, F., "An introduction to the TACOMA distributed system, version 1.0," Technical Report 95-23, University of Tromso, 1995.

.... interpreted C and 22 Tcl[65] To support state capture of a running agent, the interpreters used in the system must be able to capture their own full state (i.e. including the state of the program being interpreted) A primary drawback of these and other mobile agent systems such as TACOMA[44,45], Agent Tcl[36] and Telescript[92] is the use of interpreted execution for agents. In our intended application domain, this model fails to meet the performance requirements of most users. One system that overcomes this limitation is Extended Facile[48] an agent programming system based on the ....

D. Johansen, R. van Renesse, and F. Schneider, "An Introduction to the TACOMA Distributed System, Version 1.0," Computer Science Technical Report 95-23, University of Troms, Troms, Norway, June, 1995.

....within it. On the other hand, in Java [26] the class loader can be programmed to enable a Java program to link dynamically code downloaded from the network, hence providing support for weak mobility. Java derivatives like Aglets [15] Voyager [13] or Mole [25] as well as languages like TACOMA [12], Facile [14] and M0 [27] support weak mobility Dip. di Elettronica e Informazione, Politecnico di Milano, P.za Leonardo da Vinci 32, 20133 Milano, Italy. z Dept. of Computer Science, Washington University, Campus Box 1045, One Brookings Drive, Saint Louis, MO 63130 4899, USA. 1 by ....

D. Johansen, R. van Renesse, and F.B. Schneider. An Introduction to the TACOMA Distributed System - Version 1.0. Technical Report 95-23, Dept. of Computer Science, University of Troms and Cornell University, Troms, Norway, June 1995.

....showed up: the Software Distribution on Demand which is able not only to transport code, but also to install packages automatically. For Table 1: Mobile Agent Systems Classified Type of Mobility Systems Remote Execution Java Servlets (push [Sun96] Remote Evaluation [Stamos86] Tacoma [JovRSc95] Code on Demand ActiveX [AarAar97] Java Applets [Sun94] Java Servlets (pull [Sun96] Weak Migration Aglets [IBM96] Mole Odyssey [GenMag97] Strong Migration AgentTcl [Gray95] Ara [Peine97] Telescript [GenMag96] 3 Mobility Concepts for Mobile Agents 5 achieving that, code servers offer programs ....

D. Johansen, R. van Renesse, F. Schneider. "An Introduction to the TACOMA Distributed System - Version 1.0", Technical Report 95-23, University of Tromso, 1995.

....We argue that these different concepts and notions cannot be compared directly. It is as inappropriate and misleading as trying to compare the emacs editor, the fork UNIX system call, and the client server design paradigm. There is also confusion about terminology. For instance, several systems [17], 18] claim to be able to movethestate of a component along with its code. This assertion is justified by the availability of mechanisms that allow the programmer to pack some portion of the data space of an executing component before the component s code is sent to a remote destination. Indeed, ....

....denote a software componentthat is able to movebetween different execution environments. This definition has actually different interpretations. For example, while in Telescript [19] an agent is represented by a thread that can migrate among different nodes carrying its execution state, in TACOMA [17] agents are just code fragments associated with initialization data that can be shipped to a remote host. They do not havetheability to migrate once they have started their execution. On the other hand, in the artificial intelligence communitythe term agent denotes a software 2 ....

[Article contains additional citation context not shown here]

D. Johansen, R. van Renesse, and F.B. Schneider, "An Introduction to the TACOMA Distributed System - Version 1.0," Tech. Rep. 95-23, Dept. of Computer Science, Univ. of Troms and Cornell Univ., Troms, Norway, June 1995.

....distributed computations; and consistency, abstractions and primitives should have a strong semantics, so that programs behaviors should be predictable. In the last ten years, many agents systems have been designed, with different achievements in these three directions. Agent Tcl[Gra95] TACOMA[JRS95] and Telescript[Whi96] are examples of agents systems based on scripting languages. They achieve simplicity and consistency, thanks to a few powerful primitives to manage agents. However, they can 1 hardly be used to write complex software projects, since their scripting languages do not provide ....

Dag Johansen, Robbert Van Renesse, and Fred B. Schneider. An introduction to the tacoma distributed system version 1.0. Technical Report 95-23, University of TromsO, Norway, June 1995.

....then offer a high level interface to Prospero, which would take care of practically assigning processors. 4.3 Agents The mobile agent community deals with the problem of controlling distributed mobile computations, called agents. The most widespread agent systems are Telescript [27, 28] Tacoma [23, 24], Agent TCL [16] and Ara agents [40, 39] Very few of them are able to control resources in a similar way as Quantum. Telescript [27, 28] is an agent scripting language. Every Telescript agent has a permit that limits its capabilities in the electronic marketplace, like for instance, the right ....

Dag Johansen, Robbert van Renesse, and Fred B. Schneider. An Introduction to the TACOMA Distributed System Version 1. Technical Report 95--23, Department of Computer Science, University of Tromsl/, Norway, 1995. http://www.cs.uit.no/ DOS/Tacoma/Publications.html.

.... to evolve separately from one another without causing undefined function errors [8, 15, 25] Dynamically linked libraries partition a program into parts that can be built and changed separately [33] And mobile agent systems transfer runnable code between active entities in a distributed system [1, 2, 7, 18, 19, 29, 35]. However, these approaches alone are inappropriate for heterogeneous distributed object computing systems. Each requires a single common source language (Java, Kali Scheme, Erlang, and mobile agents) computer architecture (COM) or software environment (dynamic linking) Furthermore, with COM ....

Johansen, D., van Renesse, R., Schneider, F., "An introduction to the TACOMA distributed system, version 1.0," Technical Report 95-23, University of Tromso, 1995.

....application from scratch (using just the network services provided by the operating system) it is more convenient to use an Agent Support Environment (ASE) An ASE supplies the function that is common to many agent based applications, such as migration. Some examples of ASEs are Tacoma [7, 6, 8], Agent Tcl [4, 5] and TeleScript [9] Unfortunately, to the best of our knowledge, the current ASEs lack support for asynchronous agent to agent communication. Some do offer traditional network communication, which relies upon machine names. Therefore, the application itself has to know each ....

D. Johansen, R. van Renesse, , and F. Scheidner. An Introduction to the Tacoma Distributed System: Version 1.0. Technical Report 95-23, Department of Computer Science, University of Troms?, http://www.cs.uit.no/Lokalt/Rapporter/Reports/9523.html, 1995.

....different systems provide dissimilar services and are based on different models of computation. In order to give the reader an idea of the kind of the services provided by the ASE s, this section briefly describes three systems that have been receiving attention: Tacoma [Johansen 95a] Johansen 95b] Agent Tcl [Gray 95] Gray 96a] and Telescript [GenMagic 96] 2.1. Tacoma The Tacoma (Troms And COrnell Moving Agents) Project s main objectives are to investigate what services need to be provided in order to support easy building of agents (i.e. what exactly the ASE should be) and how ....

....as a guest program. In this system, all the agents visiting a foreign site must meet at an entry point at that site. The single entry point contains a firewall agent that can provide services like authentication, access control, accounting, and provision of fault tolerance to the guest agent [Johansen 95b] All agents arriving in a given machine are received by the firewall agent. In the current implementation of Tacoma, the firewall agent basically logs the briefcase (which includes the agent code) to disk. This means the only security service provided is accounting. In order to be executed, the ....

[Article contains additional citation context not shown here]

Dag Johansen, Robert van Renesse, and Fred Scheidner. An Introduction to the TACOMA Distributed System: Version 1.0. Technical Report 95-23, Department of Computer Science, University of Tromsø. 1995. http://www.cs.uit.no/Lokalt/Rapporter/Reports/ 9523.html.

....structures, by defining document methods for every document. As a result documents of different types can be accessed in a uniform way. 6.2. Mobile code A number of mobile code and mobile agent systems that use interpreted code have been designed and released recently, for example TACOMA [10] or the Frankfurt Mobile Agents Infrastructure (ffMAIN) 12] The ffMAIN infrastructure supports agents in virtually any implementation language and allows a tight integration with the WWW through the use of HTTP as transport protocol. Concepts used in the ffMAIN infrastructure have been adopted ....

D. Johansen, R. van Renesse, and F. B. Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical Report 95--23, June 1995.

....We argue that these different concepts and notions cannot be compared directly. It is as inappropriate and misleading as trying to compare the emacs editor, the fork UNIX system call, and the client server design paradigm. There is also confusion about terminology. For instance, several systems [17], 18] claim to be able to move the state of a component along with its code. This assertion is justified by the availability of mechanisms that allow the programmer to pack some portion of the data space of an executing component before the component s code is sent to a remote destination. ....

....a software component that is able to move between different execution environments. This definition has actually different interpretations. For example, while in Telescript [19] an agent is represented by a thread that can migrate among different nodes carrying its execution state, in TACOMA [17] agents are just code fragments associated with initialization data that can be shipped to a remote host. They do not have the ability to migrate once they have started their execution. On the other hand, in the artificial intelligence community the term agent denotes a software 2 component ....

[Article contains additional citation context not shown here]

D. Johansen, R. van Renesse, and F.B. Schneider, "An Introduction to the TACOMA Distributed System - Version 1.0," Tech. Rep. 95-23, Dept. of Computer Science, Univ. of Tromsø and Cornell Univ., Tromsø, Norway, June 1995.

....Secure Environments. Secure environments are agent based systems that are isolated, i.e. exist independent of other environments and security domains. These environments are also called ASE s and are described in Section 3. Additional information on these and similar environments can be found in [11, 23, 26, 27, 41, 47]. In an agent based environment based on the Internet, ASE s and secure co processor systems may exist as part of the infrastructure. An agent outside of such a security domain, may be able to access such domains as part of the entire system. Much work has been done to make these systems secure. ....

D. Johansen, R. van Renesse, and F.B. Schneider. An introduction to the TACOMA distributed system: Version 1.0. In University of Tromso: Technical Report 95-23, 1995.

....This included security issues like host integrity and agent integrity. To enforce security policies and to support heterogeneity, agent systems commonly use safe and secure languages and virtual machines. Chapter 3 The Tacoma system In this chapter we introduce the Tacoma model as described in [JvRS95a, JvRS95b] We present the key abstractions and services needed, and the application programmers interface (API) We will also briefly investigate other important topics like security and fault tolerance. 3.1 Introduction We begin with our discussion of the Tacoma system, by introducing basic ....

D. Johansen, R. van Renesse, and F. B. Schneider. An Introduction to the TACOMA Distributed System --- Version 1. Technial report TR-95-23, University of Tromsø, Norway, June 1995.

....was centered around the design of a special purpose language for expressing agents behaviors, including their ability to move themselves around the Internet. More recent approaches rely on existing languages, such as Java (used by IBM Aglets [LO] or Tcl Tk (used by Agent Tcl [Gra96] and Tacoma [JvS95]) They focus on the abstractions provided for agent development and use, and the supporting infrastructure, for example, by utilizing existing communication protocols like HTTP [LDD95] Another focus of mobile agents research has been on intelligent communication. This is best represented by the ....

D. Johansen, R. van Renesse, and F. B. Schneider. An introduction to the TACOMA distributed system version 1.0. Technical Report 05-23, Department of Computer Science, University of Tromsø, June 1995. http://www.cs.uit.no/DOS/Tacoma/index.html.

....ffl management of asynchronous tasks; ffl reduction in communication costs. There are several code mobility modes: 1. Remote Execution and Code on Demand: the agent is transferred before execution of the associated code so both code and data are transferred. This scheme is followed by Tacoma [6], whereas the Code on Demand scheme is used to manage Java Applets [4] 2. Weak Migration: only code state of data are transferred. This is the scheme used by Aglets and Mole. 3. Strong Migration: code data state of execution are transferred. This scheme is the most complete one and is used ....

D. Johansen, R. van Renesse, and F. B. Schneder. An Introduction to TACOMA Distributed System Version 1.0. Technical Report 95-23, University of Tromso, 1995.

....the whole UNIX process containing the interpreter is migrated, instead of a single thread within it. On the other hand, in Java [19] the class loader can be programmed to enable a Java program to link dynamically code downloaded from the network, hence providing support for weak mobility. TACOMA [10], Facile [20, 11] and M0 [21] are examples of languages that support weak mobility by allowing a procedure or function to be sent to another node for remote execution, with the portion of the global environment that is needed to proceed with execution but with no execution state. By and large, ....

Dag Johansen, Robbert van Renesse, and Fred B. Schneider. An Introduction to the TACOMA Distributed System - Version 1.0. Technical Report 95-23, University of Tromsø and Cornell University, June 1995.

....it. On the other hand, in Java [31] the class loader can be programmed to enable a Java program to link dynamically code downloaded from the network, hence providing support for weak mobility. Java derivatives like Odyssey, Aglets, Voyager [17] or Mole [30] as well as languages like TACOMA [15], Facile [18] and M0 [32] support weak mobility by allowing a procedure or function to be sent to another node for remote execution, with the portion of the global environment that is needed to proceed with execution but with no execution state. Dip. di Automatica e Informatica, Politecnico ....

D. Johansen, R. van Renesse, and F.B. Schneider. An Introduction to the TACOMA Distributed System - Version 1.0. Technical Report 95-23, Dept. of Computer Science, University of Tromsø and Cornell University, Tromsø, Norway, June 1995.

....programmer and makes agent programs more complicated. The following table classifies existing agent systems with regard to the degree of mobility supported. Table 1: Mobile Agent Systems Classified Systems Remote Execution Java Servlets (push, Sun, 1996) Remote Evaluation (Stamos, 1986) Tacoma (Johansen, van Renesse and Schneider, 1995) Code on Demand ActiveX (Microsoft, 1996) Java Applets (Sun, 1994) Java Servlets (pull, Sun, 1996) Weak Migration Aglets (IBM, 1996) Mole (Stra er, Baumann and Hohl, 1997) Odyssey (General Magic, 1997) Strong Migration AgentTcl (Gray, 1996) Ara (Peine, 1997) Telescript (General Magic, 1996) 3 ....

Johansen, D., van Renesse, R. and Schneider, F. (1995) An Introduction to the TACOMA Distributed System - Version 1.0. Technical Report 95-23, University of Tromso.

....protocol image languages and how they must behave in order to fully service agents. The AgentTCL project aimed to reduce agentmigration to a single TCL command however, there is no inter agent communication in AgentTCL. SHOE [LUKE97] is a non mobile ontology based web agent. The TACOMA project [TACO95] and Telescript [WHIT94] both focus on O.S. support for agents but do not model agent ontologies for communication. The Federation for Intelligent Physical Agents [FIPA97] and the Agent Society [AGEN97] are two emerging agent standards bodies. Conclusions and Implementation In this paper we have ....

Johansen D., van Renesse R., Schneider F., An Introduction to the TACOMA Distributed System Version 1.0, http://dslab3.cs.uit.no:1080/Tacoma/index.html, 1995

....of the various computation modules is fixed; adaptation is achieved by changing the way in which the network is used. Several systems have been built which permit an executing program to move while it is in execution for example Obliq [3] Agent TCL [11] Emerald [13] Telescript [23] and TACOMA [12].The primary distinction between these systems and Sumatra is that in Sumatra, all communication and migration happens under application control. Complete application control allows us to easily explore different policy alternatives for resource monitoring and for adapting to variations in ....

D. Johansen, R. van Renesse, and F. Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical Report 95-23, University of Tromso, 1995.

....70s, when it was used for remote job submissions [1] and that has always been available to UNIX users by means of the rsh facility. A formal definition of the remote evaluation mechanism is presented in [20] Remote code execution represents the basis for several mobile code systems like TACOMA [13], Obliq [3] and M0 [23] Suppose now that user A wants to execute code C on site B. Then, A sends B the following message : A m0 Gamma B : A; B p (C) A s (H(C) B; A 0 ; t A ; i A ) Since the code is protected using the public key of B, it is accessible by B only. When B receives the ....

Dag Johansen, Robbert van Renesse, and Fred B. Schneider. An Introduction to the TACOMA Distributed System - Version 1.0. Technical Report 95-23, Department of Computer Science, University of Tromsø and Cornell University, Tromsø, Norway, June 1995.

....of code and threads addressed by Code. However, a package providing the appropriate mechanisms on top of Code is currently under development. In Code, the unit of migration is the group. A group is simply a container for classes and objects. This abstraction is reminiscent of TACOMA briefcases [7] or, more closely, of the object group abstraction found in Sumatra [1] However, there at least two significant differences: In Sumatra, the classes needed by objects in an object group are not transferred along with the group, and must be shipped explicitly with a separate downloadClass ....

D. Johansen, R. van Renesse, and F. B. Schneider. An Introduction to the TACOMA Distributed System---Version 1.0. Technical Report 95-23, Tromsø and Cornell Univ., June 1995.

....Java [13] Other systems like Avalon [11] NCL [12] REV [17] and Obliq [10] allow programs in execution to initiate computation on remote nodes and wait for their completion. The most sophisticated support is provided by systems like Agent TCL [14] Emerald [16] Mole [22] Aglets [20] TACOMA [15] and Telescript [24] which permit an executing program to move while it is in execution. Programs that use mobility as a mechanism to adapt to resource changes have three requirements that are not shared with other mobile programs. First, they need to be aware of their execution environment. In ....

....our initial experience with a prototypical application that combines and composes weather images from multiple geographically distributed sources. In this paper, we assume that the reader is familiar with mobile code languages. For introduction to mobile code languages and mobility mechanisms, see [4, 7, 10, 11, 12, 13, 14, 15, 16, 20, 17, 22, 24]. 2 Design constraints In this section, we discuss the language design constraints that arise from the desire to be resource aware and the use of mobility as a mechanism to adapt to changes in resource availability. We discuss the requirements of awareness, agility and authority and the ....

[Article contains additional citation context not shown here]

D. Johansen, R. van Renesse, and F. Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical Report 95-23, University of Tromso, 1995.

....rather than a single thread. In turn, Java [14] is an example of a weak MCL that supports directly the COD paradigm. The Java class loader, responsible for resolving class names at run time, can be modified by the programmer to retrieve the corresponding code from a remote host. Finally, TACOMA [9], Facile [11] M0 [17] and Mole [5] are weak MCLs that support the REV paradigm. In these MCLs, a procedure or function can be sent to a remote host, at most with the portion of the global environment needed for their remote execution. 4 Network Management and Code Mobility Put Together In this ....

D. Johansen, R. van Renesse, and F.B. Schneider. An Introduction to the TACOMA Distributed System - Version 1.0. Technical Report 95-23, University of Tromsø and Cornell University, June 1995.

....1: The system architecture 2 Related Work Our work blends ideas from two areas, agents and RPC, into a single system. In this section we describe related work in these areas. The advantages of transportable agents have led to a flurry of implementation work. Three notable systems are Tacoma [JvRS95], Telescript [Whi94] and Agent Tcl [Gra95a, Gra95b] Tacoma agents are written in Tcl Horus, which is a version of Tcl that uses Horus [vRHB94] to provide group communication and fault tolerance. Tacoma agents communicate using the meet operation; data to be exchanged is carried in a briefcase, ....

D. Johansen, R. van Renesse, and F.B. Schneider. An introduction to the TACOMA distributed system version 1.0. Technical Report 95-23, University of Tromsø, June 1995.

....to initiate computation on remote nodes and wait for their completion. The most sophisticated support is provided by systems like Agent This research was supported by ARPA under contract #F19628 94 C 0057, Syracuse subcontract #353 1427 TCL [15] Emerald [17] Mole [24] Aglets [20] TACOMA [16] and Telescript [26] which permit an executing program to move while it is in execution. Programs that use mobility as a mechanism to adapt to resource changes have three requirements that are not shared with other mobile programs. First, they need to be aware of their execution environment. In ....

....mechanism which allows programs to execute a procedure at a specified site. Most mobile code languages have selected one of these two alternatives. Agent Tcl [15] Telescript [26] and Aglets [20] use a go based mechanism whereas Obliq [9] Avalon [11] NCL [13] REV [18] and TACOMA [16] use a function call based interface. To ensure a prompt response to asynchronous events, a function call based interface would require one of two things: 1) either the language automatically captures the continuation at the point at which an event occurs and makes it available for use within the ....

D. Johansen, R. van Renesse, and F. Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical Report 95-23, University of Tromso, 1995.

....Specifically, the need to support existing data mining applications with little or no modification, requires that the agent system supports different implementation languages and mechanisms that enables agents to move nonmobile software objects between sites. The first mobile agent systems [14, 8, 3] preceded the Java programming language. However, since its introduction, most, if not all new mobile agent systems have been built using Java and its associated Java Virtual Machine (JVM) Many systems of this kind are publicly known, but in essence, they provide the same functionality; move a ....

D. Johansen, R. van Renesse, and F. B. Schneider. An Introduction to the TACOMA Distributed System --- Version 1. Technial report TR-95-23, University of Troms, Norway, June 1995.

....and authorization required by the system administration policies. For example, one can implement an extension that downloads other extensions that originate from a system administrator and that are digitally signed by the corporate security ocer. Readers familiar with the original TACOMA [14] systems should be aware that the structure of TOS is di erent. In designing TOS, we have concentrated on distilling the minimal set of functions 3 Download Mobility Executor Messenger Firewall Java Virtual Machine thread Shell Carrier begin( inbox extension Figure 1: TOS ....

Dag Johansen, Robbert van Renesse, and Fred B. Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical Report 95-23, Department of Computer Science, University of Troms, 1995.

No context found.

D. Johansen, R. V. Renesse, and F. B. Schneider. An introduction to the tacoma distributed system version 1.0. Technical Report 95-23, University of TromsO, Norway, June 1995.

No context found.

D. Johansen, R. van Renesse, and F. B. Schneider. An Introduction to the TACOMA Distributed System---Version 1.0. Technical Report 95-23, Institute of Mathematical and Physical Sciences, University of Troms, Troms, Norway, June 1995.

No context found.

Dag Johansen, Robbert van Renesse, , and Fred B. Schneider. An introduction to the TACOMA distributed system version 1.0. Technical Report 95-23, Department of Computer Science, University of Troms, June 1995.

No context found.

D. Johansen, R. van Renesse, and F.B. Schneider. An introduction to the TACOMA distributed system version 1.0. Technical Report 95-23, University of Troms, June 1995.

No context found.

D. Johansen, R. van Renesse, and F. Schneider. An Introduction to the TACOMA Distributed System Version 1.0. Technical Report 95-23, UniversityofTromso, 1995.

No context found.

Dag Johansen, Robbert van Renesse, and Fred B. Schneider, June 1995a. An introduction to the TACOMA distributed system Version 1.0. Technical Report 95--23, Department of Computer Science, University of Tromsø, Norway.

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