Lockhart, H., OSF DCE: A Guide to Developing Distributed Applications, McGraw-Hill, Apr. 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....ensuring that the involved thread are monitored as acting on behalf of the same activity. An invocation of an out of process object is implemented by a remote procedure call (RPC) that is executed between the stub object and the actual object. The RPC protocol used by DCOM basically is DCE RPC [Loc94] with the extension that some extra information is piggybacked on the RPC call reply messages. Since its latest release DCOM defines in its API the ChannelHook interface that allows defining and extracting the extra information piggybacked to the RPC call reply messages [Edd99] By default, DCOM ....

Lockhart, H., OSF DCE: A Guide to Developing Distributed Applications, McGraw-Hill, Apr. 1994.

....Registry mechanism, among other reasons. DCOM also lacks a good security concept [6] other than what Windows NT provides at the operating system level. DCOM will not have any additional security functionality until NT 5.0, which claims to be using Kerberos, is released. 2. 5 Messaging Messaging [11] is another popular paradigm for distributed middleware. In this model, a software component makes its service available by publicizing the existence of a message queue that receives service requests. Clients send their requests to the service queue, whereupon the server retrieves and processes ....

H. W. Lockhart. OSF DCE Guide to Developing Distributed Applications. McGraw-Hill, 1994.

....access level and degree of transparency for distributed applications to communication functionality. In general, they may cover transaction based applications, directory services, location independent services, or dynamic object invocation. Examples of some general purpose middleware comprise DCE [1], CORBA [2] TINA C [3] COM [4] or ANSA [5] The views of these middleware approaches focus mainly on the interoperability issue as well as the generic service provision, but they do not concentrate on efficient communication protocol processing or multimedia Quality of Service (QoS) support in ....

H. W. Lockhart, OSF DCE---Guide to Developing Distributed Applications, McGraw-Hill, New York, U.S.A., 1994.

....due to at least two prevailing characteristics, namely the static monolithic implementation model, and the lack of evolution management strategies. CHAPTER 1. INTRODUCTION 8 1.2. 1 Static Monolithic Implementation Model In current distributed computing systems (e.g. Legion[32] CORBA [57] DCE [50], Mentat [28] PVM [68] MPI [52] the machine code that comprises object implementations exists in binary executable files. The mechanism that creates or starts an object, process, or task simply executes the appropriate binary. In this implementation model, changing any part of the object s ....

....of dynamically configurable distributed objects. This is because existing systems almost exclusively use static monolithic binary executables to create processes or tasks. For example, CORBA (Common Object Request Broker Architecture) 57] and DCE (Distributed Computing Environment) [50] both use interface definition language (IDL) compilers to generate high level language stubs for an object s functions; the stubs are then filled in by programmers and compiled together into a single binary executable that implements the object. 1 Like CORBA and DCE, PVM (Parallel Virtual ....

Lockhart, Jr., H.W., OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, Inc. New York 1994.

....ensuring that the involved thread are monitored as acting on behalf of the same activity. An invocation of an out of process object is implemented by a remote procedure call (RPC) that is executed between the stub object and the actual object. The RPC protocol used by DCOM basically is DCE RPC [10] with the extension that some extra information is piggy backed on the RPC call reply messages. Since its latest release DCOM defines in its API the ChannelHook interface that allows defining and extracting the extra information piggy backed to the RPC call reply messages [3] By default, DCOM ....

H. Lockhart: "OSF DCE: A Guide to Developing Distributed Applications", McGraw-Hill Inc., April 1994.

....session as modelled in Figure 8. A future issue to this work is its application to the management of a real videoconferencing application, the implementation of which is underway over a distribution platform, namely OSF DCE (Open Software Foundation s Distributed Computing Environment) [Lock94]. That integration will hold in a project called OAMS (Open management Architecture for Multimedia Services over ATM) led within our laboratory. 6 ACKNOWLEDGEMENTS The authors would like to acknowledge Prof. Noemie Simoni, from ENST, Paris, France, and Prof. Tuncay Saydam, from University of ....

H.W. Lockhart, Jr, OSF DCE : Guide to Developing Distributed Applications, McGraw-Hill, Inc., 1994.

....is a distributed processing environment (DPE) that adopts the concepts of the RM ODP computational and engineering model. The basis for the TINA DPE is a CORBA ORB. The Open Group s Distributed Computing Environment (DCE) Digital Equipment Corporation, 1992) Open Software Foundation, 1992) (Lockhart, 1994), Schill, 1996) provides a communication network and middleware platform independent platform for distributed processing systems. It is based on a clientserver architecture and does not support object oriented technologies. The main concept which should ensure compatibility, interoperability and ....

Lockhart, H.W. (1994) OSF DCE -- Guide to Developing Distributed Applications. McGraw-Hill, New York, U.S.A.

....the workstations of a cluster. However, a network of dozens or hundreds of workstations is much more powerful than a multiprocessor workstation with up to four processors, provided that the algorithm is appropriate and a distributed environment is available. Several distributed systems (e.g. DCE [5]) or software packages (e.g. MPI [3, 8] PVM [7] Linda [1] allow networked computers to appear as a single concurrent computational resource. As a matter of fact, all these programming environments require to learn a new programming model. DCE supports threads and remote procedure calls, PVM is ....

....mutual exclusion during this process. The main thread waits for the end of all five threads, afterwards the result could be used for further computations. 2 int m1[200] 200] m2[200] 200] m3[200] 200] int row=0, col=0; pthread mutex t lock; protects row and col pthread t worker threads[5]; void worker( void dummy ) int myrow, mycol, i, result; while (pthread mutex lock( lock ) row 200 ) myrow=row; mycol=col ; if (col = 200) col = 0; row ; pthread mutex unlock( lock ) result = 0; for ( i=0; i 200; i ) result = m1) myrow] i] m2) i] mycol] ....

[Article contains additional citation context not shown here]

H. W. Lockhart Jr. OSF DCE Guide to Developing Distributed Applications. McGraw-Hill, Inc., 1994.

....required technology from scratch. A large body of relevant research in distributed systems, parallel computing, fault tolerance, management of workstation farms, and pioneering wide area parallel processing projects, provide a strong foundation on which to build. Related efforts such as OSF DCE [16] and CORBA [2] are rapidly becoming industry standards. Legion and DCE share many of the same objectives, and draw upon the same heterogeneous distributed computing literature for inspiration. Consequently, both projects use many of the same techniques, e.g. an object based architecture and ....

H.W. Lockhart, Jr., "OSF DCE Guide to Developing Distributed Applications," McGraw-Hill, Inc. New York 1994.

....support and interoperability: Legion must be able to support the integration and interoperability of application components written in a variety of source languages. We feel that interoperability also dictates that we support legacy codes and work with emerging standards such as CORBA [31] and DCE [26], wherever possible. Fault tolerance: In a large scale metasystem, resource failures (hosts, communication links, disks, etc. will be commonplace. Therefore, the Legion system itself must deal with failures, through system object fault tolerance and dynamic system reconfiguration, as well as ....

.... interface Page 19 [32] Ice T [14] and Harness [8] are all outgrowths of the significant existing work in first generation network parallel computing systems, such as PVM [13] and MPI [19] and in modern transparent distributed computing systems, such as the Berkeley NOW project [1] and DCE [26]. 6.1 Globus The Globus project [12] at Argonne National Laboratory and the University of Southern California, and Legion share a common base of target environments, technical objectives, and target end users, as well as a number of similar design features. For example, similar to Legion s use ....

Lockhart, Jr., H.W., OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, Inc. New York 1994.

....2. Related work Current distributed computing systems do not support mechanisms that have the same goals and characteristics as dynamic configurability. This is because existing systems almost exclusively use static executables to create processes or tasks. For example, CORBA [28] and DCE [23] both use IDL compilers to generate high level language function stubs, which are then filled in by programmers and compiled together into a single binary executable. PVM [32] and MPI [24] create tasks by running executables from system or user bin directories. Likewise, Mentat [16] objects are ....

Lockhart, Jr., H.W., OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, Inc. New York 1994.

....must therefore be possible to integrate heterogeneous sourcelanguage application components in much the same manner that heterogeneous architectures are integrated. Interoperability also means that we must be able to support legacy codes and work with emerging standards such as CORBA [32] and DCE [28]. Fault tolerance: In a system as large as Legion it is certain that at any given instant several hosts, communication links, and disks will have failed. Thus, dealing with failure and with dynamic re configuration is a necessity for both Legion system level objects and the applications they ....

.... projects such as MOL [33] Ice T [15] and Harness [9] are all outgrowths of the significant existing work in first generation network parallel computing systems, such as PVM [14] and MPI [21] and in modern transparent distributed computing systems, such as the Berkeley NOW project [1] and DCE [28]. Page 48 8.1 Globus The Globus project [13] at Argonne National Laboratory and the University of Southern California shares with Legion a common base of target environments, technical objectives, and target end users. Beyond a basic similarity in goals, Globus and Legion also share a number ....

Lockhart, Jr., H.W., OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, Inc. New York 1994.

....data is encoded and packed into a buffer before it is sent. The data is unpacked and decoded after it is received by the receiving process. In this paper, we describe some efforts to improve the performance of PVM s packing unpacking functions by introducing the receiver makes it right (RMR) [5] working mechanism. The current version of PVM supports three different methods of message packing: PvmDataRaw, PvmDataInPlace, and PvmDataDefault. PvmDataRaw does no encoding or decoding. The raw data is sent between PVM tasks. PvmDataInPlace also uses raw data with the added feature that no ....

....(like Sun Microsystems s ONC RPC) facility is at the heart of OSF DCE. DCE RPC and PVM have some goals in common. One of the services DCE RPC provides is to transport data across the network and convert them from one format to another as needed. DCE RPC is based on the Apollo Computer s NCS scheme [5, 10]. NCS (Network Computing System) has some advantages, but data formatting was not one of them. Instead of XDR, NCS had what its authors called receiver makes it right . In XDR, everyone converts to a specified format and some machines pay a penalty. In NCS, everyone can convert to one of a couple ....

H. Lockhart, Jr., OSF DCE: Guide to Developing Distributed Applications, McGraw-Hill, Inc., 1994.

....an efficient name resolution in such condition. As the number of objects useful for the collaborative application is small compared to the number of objects potentially manipulated by the underlying world wide distributed system, it is inefficient to rely on a global name service such as DCE[Loc94] Indeed, the architecture of the underlying network makes the name services of DCE cumbersome to use since it is decomposed into two entities: a cell directory service and a global name service. The first manages the name within its corresponding cell and the second manages the cell name. With a ....

....a unique contextual name, thus it is not necessary to perform a detection of conflicting names. This ensures a better locality and scalability for the registration since it is not necessary to verify all the entries maintained by the name service at each registration. Name service (e.g. DCE s[Loc94] X500[CCI88] suffers from the impossibility to disable this detection. The way that the contextual name is used by existing name services is also a lack of flexibility. Systems like Propero[Neu92] DCE[Loc94] X500[CCI88] DNS[TPRZ84] are hierarchical name servers that use the format of the ....

[Article contains additional citation context not shown here]

H.W. Lockhart. OSF DCE Guide to Developing Distributed Applications. Mc GrawHill, 1994.

....with multimedia databases and end to end QoS guarantees. By providing BeeHive wrappers to Legion we will be able to support Legion objects within BeeHive that satisfy QoS guarantees. Commercial systems with similar goals as Legion, however, focused exclusively on a client server model, are OSF DCE [40] and CORBA [63] The CORBA standards and products based on them also do not have the functionality nor real time properties that we are developing, although a real time CORBA is emerging. In recent years, considerable progress has been made in the areas of QoS support for operating systems, ....

H. W. Lockhart, OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, New York, 1994.

....required technology from scratch. A large body of relevant research in distributed systems, parallel computing, fault tolerance, management of workstation farms, and pioneering wide area parallel processing projects, provide a strong foundation on which to build. Related efforts such as OSF DCE [9] and CORBA [2] are rapidly becoming industry standards. Legion and DCE share many of the same objectives, and draw upon the same heterogeneous distributed computing literature for inspiration. Consequently, both projects use many of the same techniques, e.g. an object based architecture and ....

H.W. Lockhart, Jr., OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, Inc. New York 1994.

....are known by their interfaces. A module interface is specified in the system s object oriented interface definitions language. In most systems, Interface Definitions Languages have been designed to facilitate such description (e.g. as Xerox ILU, OMG ORB, and OSF DCE Remote Procedure Call (RPC) [18]) 2.1 CORBA: A Brief Overview To narrow down its scope to an implementation level, this paper relates directly to the Common Object Request Broker Architecture (CORBA) often known as The CORBA Standard [21] In particular the selected CORBA implementation is the IONA implementation of CORBA. ....

H. W. Lockhart, "OSF DCE: Guide to Developing Distributed Applications"; McGraw-Hill, 1994.

....required technology from scratch. A large body of relevant research in distributed systems, parallel computing, fault tolerance, management of workstation farms, and pioneering wide area parallel processing projects, provide a strong foundation on which to build. Related efforts such as OSF DCE [7] and CORBA [2] are rapidly becoming industry standards. Legion and DCE share many of the same objectives, and draw upon the same heterogeneous distributed computing literature for inspiration. Consequently, both projects use many of the same techniques, e.g. an objectbased architecture and model, ....

H.W. Lockhart, Jr., OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, Inc. New York 1994.

....with multimedia databases and end to end QoS guarantees. By providing BeeHive wrappers to Legion we will be able to support Legion objects within BeeHive that satisfy QoS guarantees. Commercial systems with similar goals as Legion, however, focused exclusively on a client server model, are OSF DCE [40] and CORBA [64] The CORBA standards and products based on them also do not have the functionality nor real time properties that we are developing, although a real time CORBA is emerging. In recent years, considerable progress has been made in the areas of QoS support for operating systems, ....

H. W. Lockhart, OSF DCE Guide to Developing Distributed Applications, McGraw-Hill, New York, 1994.

....an efficient name resolution in such a condition. As the number of objects useful for the collaborative application is small compared to the number of objects potentially manipulated by the underlying world wide distributed system, it is inefficient to rely on a global name service such as DCE [Loc94] Indeed, the architecture of the underlying network makes the name services of DCE cumbersome to use since it is decomposed into two entities: a cell directory service and a global name service. The first manages the name within its corresponding cell and the second manages the cell name. With a ....

....unique contextual name, thus it is not necessary to perform a detection of conflicting names. This ensures a better locality and scalability for the registration since it is not necessary to verify all the entries maintained by the name service at each registration. Most name services (e.g. DCE s [Loc94] X500 [CCI88] suffer from the impossibility to disable this detection. The way that the contextual name is used by existing name services is also characterized by a lack of flexibility. Systems like Propero [Neu92] DCE [Loc94] X500 [CCI88] DNS [TPRZ84] are hierarchical name servers that use ....

[Article contains additional citation context not shown here]

H.W. Lockhart. OSF DCE Guide to Developing Distributed Applications. Mc Graw-Hill, 1994.

....power of small computers and inherent fault tolerance associated with the distribution of computing resources. 1 This project was supported by IBM Austin under contract number 32525 42640 CS. From the programming point of view, distributed processing can be categorized into various models[1]. Here we explain the client server model, the model used in this paper. For a discussion of other models refer[1] In the client server model, each interaction is characterized by an asymmetrical relationship between two software processes. One, the client, makes requests for service and the ....

....1 This project was supported by IBM Austin under contract number 32525 42640 CS. From the programming point of view, distributed processing can be categorized into various models[1] Here we explain the client server model, the model used in this paper. For a discussion of other models refer[1]. In the client server model, each interaction is characterized by an asymmetrical relationship between two software processes. One, the client, makes requests for service and the other, the server, provides the service on request. The processing flow is characteristically synchronous(blocking) in ....

[Article contains additional citation context not shown here]

H. W. Lockhart, "OSF DCE, Guide to Developing Distributed Applications," McGraw Hill, Inc., 1994.

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