B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the 15th ACM Symposiom on Operating Systems Principles, pages 68--78, December 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....but in a significantly more restricted fashion than in InterWeave. Pointers and recursive types were not supported, and all shared data had to be accessed indirectly through a local mapping table. Systems that support process or thread migration among heterogeneous computers such as Emerald [13] and Tui [11] employ data marshaling for mobility, but they do not employ data caching. As a result, they do not face the challenges of representing diffs in wire format. They also depend on compiler support to extract data type information. Smart RPC [7] is an extension to conventional RPC that ....

B. Steensgaard and E. Jul. Object and Native Code Thread Mobility among Heterogeneous Computers. In Proc. of the 15th ACM Symp. on Operating Systems Principles, pages 68--77, Copper Mountain, CO, Dec. 1995.

....Among them, Delirium [13] is the only one adopting a shared memory programming model and it requires that users 18 list explicitly all variables that each routine might destructively modify. Systems that support process or thread migration among heterogeneous computers such as Emerald [18] and Tui [16] employ data marshaling for mobility, but they usually do not employ data caching. As a result, they do not face the challenges of representing diffs in wire format. They also depend on compiler support to extract data type information. Smart RPC [9] is an extension to conventional ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the 15th ACM Symposium on Operating Systems Principles, pages 68--77, Dec. 1995.

....PB3A follow the same general scheme, thereby avoiding the centralized naming bottleneck and vulnerability. In addition to drawing upon distributed operating systems, the PB3A resembles other mobile object programming environments. One such system is the Emerald Distributed Programming Language [5] developed at the University of Copenhagen. The Emerald project created a distributed programming system for heterogeneous computer networks. This system operates in native code and native data representations on each individual platform, but marshals the data into platform independent ....

Steensgaard, Bjarne and Eric Jul. "Object and Native Code Thread Mobility Among Heterogeneous Computers ", Proceedings of the 15th ACM Symposium on Operating Systems Principles, pp. 68 - 78, December 1995.

....PB3A follow the same general scheme, thereby avoiding the centralized naming bottleneck and vulnerability. In addition to drawing upon distributed operating systems, the PB3A resembles other mobile object programming environments. One such system is the Emerald Distributed Programming Language [5] developed at the University of Copenhagen. The Emerald project created a distributed programming system for heterogeneous computer networks. This system operates in native code and native data representations on each individual platform, but marshals the data into platform independent ....

Steensgaard, Bjarne and Eric Jul. "Object and Native Code Thread Mobility Among Heterogeneous Computers", Proceedings of the 15th ACM Symposium on Operating Systems Principles, pp. 68 - 78, December 1995.

....universe. Most mobile agent platforms provide such functionality, e.g. IBM s Aglets [15] Mole [4] or Concordia [31] Moving processes. The system should allow process migration, allowing a thread interrupted on one machine to resume its execution at the same point on another machine. Emerald [17], Sprite [10]and Charlotte [3] are distributed systems providing process migration facilities. Traditionally, computing environments are heterogeneous, consisting of a mix of di erent hardware and operating systems. The adventofJava [26] which provides the abstraction of a homogeneous ....

Jul, E., Steensgaard, B., Object and Native Code Thread Mobility among Heterogeneous Computers, ########### ## ### #### ######### ## ######### ####### ########## #########, Copper Mountain Resort, Colorado, USA, December 1995. http://www-ece.rice.edu/SOSP15/paper-list.html

.... the configuration of distributed systems [1, 3, 14] A significant number of projects have explored migration in distributed systems [33] Notable examples include migration at the operating system level, as provided by Sprite [16] and at the programming language level, as provided by Emerald [27, 39]. In these systems, providing support for a uniform execution environment across all nodes and migration of application and execution state has resulted in considerable complexity. In contrast, many mobile agent systems, such as IBM s aglets [29] avoid this complexity by implementing what we call ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the 15th ACM Symposium on Operating Systems Principles, pages 68--77, Dec. 1995.

....of code. Omniware [2] Safe TCL, and Java allow programs to download code from a remote site and execute it locally. Avalon Common Lisp, REV, NCL [21] and Obliq allow programs to send code to a remote site, and receive the results of the subsequent computation. Finally, Agent TCL [27] Emerald [66], Mole [67] Aglets [47] TACOMA, and Telescript allow programs to move themselves from one node to another. The DCDO model does not technically specify the format of implementation components different implementations could have different styles of representation. However, as described ....

Steensgaard, B., Jul, E., "Object and native code thread mobility among heterogeneous computers," 15th ACM Symposium on Operating Systems Principles (SOSP), December 1995.

....support this specialized use of subroutine call and return mechanisms involve the introduction of poll points and subroutine prologues. Poll points are introduced points in execution at which the process determines if a checkpoint should be produced (analogous to Bus Stops in Heterogeneous Emerald[82]) At a poll point, if 10 the process determines that a checkpoint should be produced, code is executed to save the state of the current active subroutine and to return immediately to the calling subroutine. Poll points are placed throughout the program in a manner that ensures that after any ....

....code program) essentially, the process effects its own restart. Our approach extends this desirable feature of autonomy to include state capture as well as state restore. A more recent and fully implemented approach to the heterogeneous state capture problem was presented by Steensgaard and Jul[82], who developed an extension of the thread and objectmobility capability of the heterogeneous Emerald distributed system to allow native code migration among heterogeneous hosts (previous implementations supported native code mobility for homogeneous hosts, plus heterogeneous mobility for ....

B. Steensgaard and E. Jul, "Object and Native Code Thread Mobility Among Heterogeneous Computers," in Proceedings of the Fifteenth ACM Symposium on Operating Systems Principles, December, 1995.

....others. 5. Related work Process migration techniques have been investigated in depth [7 9] Recent work has investigated using heterogeneous migration in conjunction with typesafe languages, and in situations where the type safety of applications written in non typesafe languages can be verified [10]. Harmony is different in that we focus on the policy issues of when to migrate large, long running distributed applications; and whether to migrate the process or the data. Farming computational objects to nodes of a distributed system has been exploited by many projects [1119 ] Several ....

B. Steensgaard and E. Jul, "Object and Native Code Thread Mobility Among Heterogeneous Computers," in Proceedings of the 15 th ACM Symposium on Operating Systems Principles, 1995.

....site, preserving its execution states such as the call stack and a part of the heap image. The thread at the departure site will vanish and an equivalent thread will appear at the destination. In this sense, our model of migration is similar to those of Arachne threads system [5] and Emerald [16] rather than major Java based mobile agent systems. In our system, a place or a location to which a mobile agent migrates is a Java virtual machine. A mobile agent (thread) hops around a group of Java virtual machines. Basically, a heap image that a migrating thread can refer to is duplicated to ....

B. Steensgaard and E. Jul. Object and Native Code Thread Mobility among Heterogeneous Computers. In Proceedings of the ACM Symposium on Operating Systems Principles, pages 68--78, 1995.

....the other hand, they typically allow the migration of compiled code, which no current mobile agent systems support. Notable recent examples include Sprite [DO91] and Condor [LS92] both of which allow runtime migration of an entire process within a cluster of homogeneous machines; Emerald [JLHB88, SJ95] which supports object migration within a heterogeneous cluster, if the object code is pre installed; Legion [GWtLT97] which provides runtime migration within a heterogeneous cluster, again if the code is pre installed; and several Javabased systems [AAB98, RN98] which distribute parallel ....

Bjarne Steensgaard and Eric Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the Fifteenth ACM Symposium on Operating Systems Principles, pages 68-78, Copper Mountain, CO, December 1995. ACM Press.

....user interface, and could have been optimised out of the process. These results show that it is possible to migrate the video player with the end user disrupted for a time equivalent to less than two frames of video. 4 Related Work. The Emerald distributed programming language has been extended [20] such that objects may be migrated between heterogeneous computers. It uses a mechanism for marshalling thread and object state in a manner similar to that described in this paper, except that there is no requirement for a separate IDL description of state. Instead, the Emerald compiler generates ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of 15th Symposium on Operating System Principles, Copper Mountain, Colorado, December 1995. ACM.

....can complement the other one (e.g. the use of agents while a platform is running in disconnected mode, following by the use of an object like entity when that platform is wired) Second, for improved performance, we will explore the execution of agents with native code. Systems such as Emerald [14] have explored native code mobility but this work does not exploit compiler technology to improve agent execution performance. Third, we will explore how the implementations of objects and agents can adapt to changing application requirements. Fourth and perhaps most importantly, we will not ....

B. Steensgaard and E. Jul. Object and Native Code Thread Mobility Among Heterogeneous Computers. In Proc. of 15th SOSP, 1995.

....about the environment. ObjectStore, like other database systems, assumes that the used environment is closed# Planet does not. Consequently, ObjectStore does not have protection policy checking functionality against malicious users or program codes. 5. 4 Handling Heterogeneity Reference [28] reports a scheme for porting the Emerald mobile object system in a heterogeneous environment. It is related to the scheme described in Section 4.1 despite the fact that Emerald s scheme is designed for a single language mobile object system. Our scheme is designed for a language neutral one. ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proc. of the 15th ACM Symposium on Operating Systems Principles, pages 68--78, 1995.

....instead. Such an environment should o er adequate object models and programming paradigms, as well as structuring concepts, to cope with the complexity of large scale applications. 5 2. 2 Language based Environments Several research projects, such as COMANDOS [Con92] GUIDE[BLR94] or Emerald [SEJ95] to name only few tried to provide homogeneous distributed programming environments by following a language based approach. The employed languages o er high level concepts to develop distributed object oriented programs. Unfortunately, all of them lack appropriate structuring features. Though ....

B. Steensgaard and E. E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the 15th ACM Symposium on Operating Systems Principles, pages 68-78, Dezember 1995.

....at runtime and the migration mechanism is more transparent to the programmer. Examples of the second approach include: HMF [21] Process Inspection [14] HiCaM [25] Ythreads [30] Arachne [8] DOME [31] Shao and Schnabel s work [32] and MpPVM [6] Examples of the third approach include: Emerald [35], Tuis [34] Shub, Dubach and Rutherford s work [9, 10] Hollander and Silberman s work [17] Distributed C [24] Theimer and Hayes work [38] and porch [36] Our work is based on the third approach, as it provides the most optimal results. With heterogeneous task migration, most research has ....

....Distributed C [24] Theimer and Hayes work [38] and porch [36] Our work is based on the third approach, as it provides the most optimal results. With heterogeneous task migration, most research has focused on how to reconstruct the task s state [38, 3, 7, 30] the location of migration points [4, 35] and analysing the safety aspects of this approach [34, 19] Very little research has been done on how to optimising migration based on different kinds of tasks and different degrees of heterogeneity. Most of the work in this area has been done by the University of Colorado at Colorado Springs ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. Operating Systems Review, 39(5), Dec. 1995.

.... There are other ways of duplicating this functionality, such as using a preprocessor to automate the task of embedding type program information into the checkpoint [32] employing restricted languages that may embed architecture independent checkpointing into the compiler and runtime system [37] or employing special checkpointable data structures [5] While prototype implementations of the above techniques have been promising, the structure of NetSolve allows us to test their use with real applications, compare their performance and deploy them. We anticipate that by exploring these ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In 15th Symposium on Operating Systems Principles, pages 68--78. ACM, December 1995.

....same object. Our support for heterogeneity is derived from Java; the same code may run on different machines. We also allow different memory spaces among machines, but the object handle tables are conceptually consistent. Thus, references (pointers) to an object are identical on all the sites. In [13], machine dependent code migrates between computers, but the cost is the requirement of supplying native code for different types of machines. JavaNOW [14] provides an easy interface for the user to create a group of processes of an application. Processes communicate in a way similar to PVM, and ....

Bjarne Steensgaard and Eric Jul. Object and Native Code Thread Mobility Among Heterogeneous Computers. In 15th ACM Symposium on OS Principles, pages 68-- 78, 1995.

....this allowed us to write an extremely light weight interpreter that processes the scripts eciently. On another hand, it limits the expressiveness of the recon guration agents. By using a more powerful language for recon guration such as OCL [BBB 98] or a general purpose language like Emerald [JS95] or Java, the administrator would be able to write more sophisticated recon guration scripts that could, for example, use the agent and ORB state to make decisions about its next actions. To achieve that, our group is extending the infrastructure to add support for Java agents. We encapsulated ....

Eric Jul and Bjarne Steensgaard. Object and Native Code Thread Mobility among Heterogeneous Computers. In Proceedings of the 15th ACM SOSP), Copper Mountain, Colorado, December 1995.

....the stack, nonshared global data, and the shared data segment(s) Registers are easily copied from one machine to another. We do not have to worry about volatile variables or compiler optimizations since migration only occurs when threads voluntarily yield the processor through a procedure call [12]. Non shared global data Non shared global data refers to heap data and to statically allocated data. D CVM only ensures consistency of data explicitly allocated through D CVM calls. Hence, heap and statically allocated data is not consistent across nodes. The result is that threads on a single ....

B. Steensgaard and E. Jul, "Object and Native Code Thread Mobility Among Heterogeneous Computers," in Proceedings of the 15th ACM Symposium on Operating Systems Principles, 1995.

....stacks at runtime is prohibitively expensive, we advocate a solution based on the use of a preprocessor [12] for C C . With such a setup, the user writes programs in that language, preprocesses the code, and finally compiles with a conventional compiler. A related effort has been reported in [15], where a group of researchers developed a heterogeneous compiler for the Emerald programming language [3] This compiler inserts instructions that encode the function s state, thus enabling threads and objects to freely migrate between the four supported architectures: Sun 3, Sun SPARC, VAX, and ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the ACM Symposium on Operating Systems Principles, pages 68--78, 1995. This article was processed using the L A T E X macro package with LLNCS style

....and compiles with a conventional compiler. Though Ythreads provides for heterogeneous thread migration, it is limited in that thread primitives can only be invoked in top level thread functions. They cannot be invoked in nested functions. Emerald A somewhat similar effort has been reported in [24], where a group of researchers rewrote an existing Emerald compiler in a way that enables heterogeneous thread migration. 2 This compiler examines the source code and inserts appropriate instructions immediately before migration requests. These instructions encode and pack a function s state. ....

....will require porting the entire compiler. Also, there are practical problems related to integrating Emerald threads and objects with software that already relies on some threads library. Finally, there is the important issue of performance. It was found that the compiler developed in the project [24] took 60 longer to migrate and subsequently invoke a function than the Emerald compiler that supports only homogeneous thread migration [9] SimCal The idea behind our design, especially the use of a code preprocessor, has its origins in work done by Malloy and Soffa [12] By augmenting Pascal ....

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the ACM Symposium on Operating Systems Principles, pages 68--78, 1995.

....user to monitor the agents that flow through their workstation, examine the resources that they use, and control resource management policies. Related work Our work builds on previous work in agents [Age94] primarily in the AI community, and in transportable code [GM94, TTP 95, JdT 95, SJ95] primarily in the systems community. There are a few other projects about transportable agents [JvS95, Whi94a, Whi94b, Col95, CGH 95, SG90] We will expand this section in the full paper. Status Agent Tcl has been publically released and is in active use at several sites in various ....

Bjarne Steensgaard and Eric Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the Fifteenth ACM Symposium on Operating Systems Principles, pages 68--78, December 1995.

No context found.

B. Steensgaard and E. Jul. Object and native code thread mobility among heterogeneous computers. In Proceedings of the 15th ACM Symposiom on Operating Systems Principles, pages 68--78, December 1995.

No context found.

B. Steensgaard and E. Jul. Object and Native Code Thread Mobility among Heterogeneous Computers. In Proc. of the 15th ACM Symp. on Operating Systems Principles, pages 68--77, Copper Mountain, CO, Dec. 1995.

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