Andrew S. Tanenbaum. Distributed Operating Systems. Prentice-Hall, Englewood Cliffs, NJ 07632, USA, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....token passing can increase the latency for servicing new request, eager token passing can increase the overhead of unnecessary token passes. As a final caveat, we note that handling lost tokens in DALA is similar to handling the failure of a coordinator within a group of distributed processes [22]. Group members can check on the status of the token holder by exchanging heartbeat messages. In the event that the token holder is down, a new token holder can be chosen by running a simple leader election algorithm within the group [22] Such a strategy fails only when the token holder was the ....

....of a coordinator within a group of distributed processes [22] Group members can check on the status of the token holder by exchanging heartbeat messages. In the event that the token holder is down, a new token holder can be chosen by running a simple leader election algorithm within the group [22]. Such a strategy fails only when the token holder was the sole member of the group (and has failed) and no other copy of the video exists in the cluster; in such cases, the video can no longer be served by the cluster until a copy is retrieved from tertiary storage. 2.4 Adapting the Group Size to ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice Hall, 1995. 7

....in the copy. Unfortunately, there has been confusion on the issue. Based on existing implementations of call by copy restore for primitive (non pointer) types, an often repeated (mistaken) assertion is that call by copy restore semantics implies that shared structure results into multiple copies [16][17] 21] 4.2 DCE RPC The DCE RPC specification [10] is the foremost example of a middleware design that tries to enable distributed programming in a way that is as natural as local programming. The most widespread DCE RPC implementation nowadays is that of Microsoft RPC, forming the base of ....

Andrew S. Tanenbaum, Distributed Operating Systems, Prentice-Hall, 1995.

....the active nature of the memory (e.g. by periodically bringing pages into memory so that they can execute) 2.2 Distributed and Parallel Systems The advent of inexpensive yet ecient networking hardware has made it possible to design systems speci cally for distributed environments. Tanenbaum [25] enumerates several advantages to distributed systems. First, it may be possible in many cases to create a cluster of very inexpensive workstations (e.g. Beowulf [24] that outperforms more expensive supercomputers. Second, some systems are inherently distributed, such as retail chain inventory ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice Hall, 1995.

....and the development of highspeed networks, distributed systems are an attractive alternative to centralized computer systems. A key reason is performance; by using multiple processors, distributed systems can provide a better price performance ratio than can a single large central machine [25]. Distributed systems provide advantages for resource and data sharing, fault tolerance, and incremental growth. They also support inherently distributed activities which are not supported by centralized uniprocessor machines or multiprocessor shared memory systems. Distributed applications are ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1995.

....of main memory; this percentage is determined at the time of system generation, and is typically set to 10 . More recent systems allow the barrier between file cache and program memory to vary, allowing file caches to grow to be virtually the full size of main memory if warranted by the workload [Tanenbaum85, Stem94]. File servers, for example, will surely use much more of their main memory for file cache than will most client workstations. Our benchmark varies the amount of data and measures performance to determine the maximum file cache size for each system. Figure 4 shows these maximum file cache sizes, ....

Andrew S. Tanenbaum and RobbertVan Renesse. Distributed Operating Systems. Computing Surveys, 17(4):419-470, December 1985.

.... multi computer, IPC, DSM, Linux, WAN, TCP IP, heterogeneous environment Introduction Considering the limitations in increasing the processing speed of uni processor systems, using multiple processing elements (PEs) to solve a problem is becoming the solution of choice among computer users [1]. Here solving a problem requires the following abilities: Starting programs on different PEs. Providing them with the needed data and collecting the processed results. Synchronizing them, so that they access the data and results in an orderly manner. There are two main trends in ....

....programs on different PEs. Providing them with the needed data and collecting the processed results. Synchronizing them, so that they access the data and results in an orderly manner. There are two main trends in designing parallel systems: Multi Processors (MPs) and MultiComputers (MCs) [1,8]. MPs use processors with access to some shared memory to do their work. This shared memory makes the programmer s job in transferring data and synchronizing between the processors easy [2,3] Usually the operating system undertakes the task of running parts of a parallel application on different ....

[Article contains additional citation context not shown here]

Andrew S. Tanenbaum, Distributed Operating Systems, Prentice-Hall, 1995.

....will come up with an informal framework covering the basics in the following. 1.1.1 Distributed applications A distributed application is a program consisting of two or more communicating tasks running in a MIMD environment. The term MIMD (Multiple Instruction stream, Multiple Data stream; see [Tan95] includes any type of architecture with one or more independent computers each with independent program counters, program and data. This could be workstations in a LAN, a hyper cube or a transputer. The definition of distributed applications makes no assumptions about the a specific ....

....tools like EBBA enables the user to focus on the actual application behavior. This approach uses traces from an application and compares these traces with models specified in a form of regular expressions. Other systems like the debugger developed for the Amoeba distributed operating system (see [Tan95] uses the same approach. Complexity of program behavior is reduced through mechanisms for filtering traces of events and model checking is implemented through recognizers. These are also specified in a form of regular expressions. In [CM91] 1.3. Present technologies 13 Figure 1.3: This ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice Hall, 1995.

....include C Linda, C Linda, Modula 2 Linda, and Fortran Linda; these implementations also consist of a run time kernel that implements interprocess communication and process management. Other languages were introduced after Linda that support the distributed data structure concept; these are Orca [52], SDL [96] and Tuple Space Smalltalk [95] 5.5.3 Linda in HTMT The HTMT is a distributed shared memory model, with significant processing in the memory. Thus, using Linda functions, the search and memory management in the PIM memories could be done in parallel at the memory, and Linda s ....

Andrew S. Tanenbaum, Distributed Operating Systems, Prentice Hall, Inc., Englewood Cliffs, NJ 07632, 1995.

....Environment, Distributed Shared Memory, Remote Procedure Call, TCP IP, Linux. Introduction to DIPC DIPC is simple a software only solution to enable easy and transparent data exchange between the processes of a distributed application [8,9] Using it, people can build and program multi computers [1], built of ordinary personal computers connected over a TCP IP network [6] The current implementation enables Linux programmers to use UNIX System V IPC mechanisms [2] including shared memories, messages and semaphores, in a distributed environment. DIPC s services are accessible via the Linux ....

Andrew S. Tanenbaum, Distributed Operating Systems, Prentice-Hall, 1995.

....reliable computing platform. The only problem is that Linux was originally based on decades old designs [7] making it less attractive for more technically minded users. Linux s answer to this problem is in either port and adaptation, or introduction of newer concepts. Building Multi Computers [1] and programming them are among the more popular research subjects, and demand for them is rapidly rising. Any solution to distributed programming under Linux should keep up with one of Linux s more important features: availability to ordinary users. Motivation Linux already has symmetric ....

....(Parallel Virtual Machine) software [8] The need for a simpler distributed programming model, usable by more programmers, is clearly present. The DIPC Software DIPC (Distributed Inter Process Communications) is a software only solution for enabling people to build and program multi computers [1] easily. Each node can be an ordinary personal computer. These nodes should be connected to each other by a TCP IP [3] network. It does not use network broadcasting, which helps it work in networks with no such capabilities. There is also no assumption of the existence of a synchronized clock. ....

[Article contains additional citation context not shown here]

Andrew S. Tanenbaum, Distributed Operating Systems, Prentice-Hall, 1995.

.... that of reactive systems [26, 78, 79] The formal specification disciplines are best understood, as we shall claim, in a semantics setting [50, 87, 103, 32, 91] Disciplines that straddle programming methodology and software engineering are typically covered in subjects such as operating systems [107, 108], database systems [30, 31, 114] distributed systems [77] and data communication including protocols [104] Students of computing science learn to construct the equations that students of computer science can solve 2.4 Software Engineering Software engineering is characterised by ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice Hall, 1994.

....by means of messages. Therefore, a DM machine is harder to use than a multiprocessor; however, it is scales much better than a multiprocessor. Consequently, the largest parallel machines are DM machines. Additionally, a DM 17 machine is generally simpler and cheaper to build than a multiprocessor [52]. A multicomputer is a collection of uniprocessor nodes that are connected by an interconnection network. A cluster of workstations is another DM machine that is very similar to the multicomputer; however, the interconnection network usually has more latency and less bandwidth. 1.2 Filaments ....

....On a DM machine, processes on different nodes communicate and synchronize with messages. Explicit message passing adds many complications to a program, such as flow control and buffering of messages. Programming with messages is tricky despite the various solutions that have been proposed [52]. Moreover, the shared variable model is the most widely used model; for example, most algorithms are expressed in terms of shared memory [46] Figure 3.2 shows how Jacobi iteration might be programmed using explicit mes 49 do maxdiff epsilon # send row up and receive from below if not ....

Andrew S. Tannebaum. Distributed Operating Systems. Prentice-Hall, Inc., 1995.

....messages and replicated copies of data can cause the same problems. These problems led to the concept of consistency models. A consistency model is a specification of the allowable behavior of memory. It can be seen as a contract between the memory implementation and the program utilizing memory [45]. The input to memory is a set of memory operations (reads and writes) partially ordered by program order. The output of memory is the collection of values returned by all read operations. A consistency model is a function that maps each input to a set of allowable outputs. The memory ....

....The original definition given by the designer of a consistency model may use notation and assumptions that are intuitively familiar to the author, but appear ambiguous to others. For example, there has been some question whether Goodman s processor consistency was intended to be stronger than PRAM [41, 45]. Later work can produce a restatement of a consistency model more intuitive to the research community at large. A common trend in the literature is the development of uniform frameworks and notation to represent models previously defined [2, 3, 7, 41] There are currently two common methods of ....

[Article contains additional citation context not shown here]

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice-Hall, Englewood Cli#s, NJ, 1995.

....framework so the model should provide ways to allow LINDA abstract I O. The LIGIA model assumes the existence of an I O tuple space which is the abstraction of Input Output within the model and is a way of hiding I O physical location from users. In a nutshell LIGIA makes I O like a capability [Tan95] that can be passed via tuple spaces; this gives the kernel knowledge of which process is able to perform I O operations. At the model level, the I O tuple space represents the capability of doing I O; from the kernel point of view if a process has an I O handle this process can do I O. Later in ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice-Hall, Englewood Cliffs, NJ 07632, USA, 1995.

.... sendmail , pid, fv, good , 152, 0) username, sendmail , pid, fv, bad , 480, 19) fv, good , 161, 36) fv, good , 2, 0) pid, fv, good , 6, #define accc fv[1406] #define aayk fv[634] #define aacp fv[67] #define aacn fv[65] #define aacb fv[53] #define aaaz fv[25] #define aaai fv[8] #define aaae fv[4] color SysCall = product string string unsigned int boolean[ Figure 17: RIPPER rules for Sendmail modeled as a CPN 18 4.3 Anomaly Detection Rules as a CPN The FTP Bounce and NFS Misuse CPN models misuse intrusion detection. To illustrate the ....

....number in the token representing event a must be less than the sequence number in the token representing event b, and no two events may have sequence numbers. Literal wall clock times used for comparisons are a problem when the times are obtained from di erent computers in a distributed system [25]. Each computer has its own notion of the current time, and computer clocks tend to skew at di erent rates. We assume that the clocks are kept synchronized by the intrusion detection system (synchronized at least once every twenty four hours) and the skew between a computer s clock and the ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice-Hall, Englewood Clis, NJ, USA, 1995.

....more than one file server contributing to the file service, each located on a different machine. This distribution of file servers should remain transparent to clients. File services can be split into two types, depending on whether they support an upload download model or a remote access model [22]. In the upload download model, whole files are transferred between client and server. The client carries out operations on the file locally, before uploading the modified file to the server. The advantage of this model is its conceptual simplicity. There is no complicated file service interface, ....

....both unauthorised and authorised users gaining access to information that should be protected from them. Authorisation is commonly implemented using Access Control Lists (ACL s) which is are a list of users associated with each file in the system, specifying who may access the file and how [22]. 4.4 Kerberos Kerberos is a distributed authentication service that allows a process (a client) running on behalf of a principal (a user) to prove its identity to a verifier (an application server) without sending data across the network that might allow an attacker or the verifier to ....

Andrew Tannenbaum. 1995. Distributed Operating Systems. Prentice Hall, international edition.

....de software e implantaci on, aspectos relacionados con lenguajes, seguridad y criptolog ia. 2 El Area y Material Existente. Hace ya bastante tiempo que existen libros y se imparten cursos relacionados a sistemas distribuidos, con un enfoque que estudia los aspectos de arquitectura, por ejemplo, [24, 25, 23]. Es decir, la organizaci on en modulos de acuerdo a su funci on. Muy comunes son los cursos con este enfoque de redes de computadoras, sistemas operativos y bases de datos distribuidas. En los ultimos diez a nos, se ha desarrollado el enfoque que sigue este curso. Un enfoque que intenta ....

Andrew S. Tanenbaum. Distributed operating systems. Prentice-Hall, 1995.

....incurred in distributed processing under the SDDS with live optimization. We want to show the following: ffl Optimal Size of a Distributed Partition ffl Overhead Added by SDDSs 5. Implementation and Performance Study 8 ffl Performance Scalability We use a network multicomputer [Cul94] Tan95] in our case a number of Silicon Graphics O 2 s, running IRIX6.3, each having 64 MBytes of memory. For communication the office network is used. This network is a mix of ATM switches and Ethernet. Each workstation has a 180 MHz, R5000 MIPS CPU. All measures are given in milli seconds (ms) ....

Andrew S. Tanenbaum. Distributed Operating Systems. 1995.

....calls. These system calls are obviously location dependent, i.e. the output of these calls depend upon the node on which these are executed. The system times of two different machines in a distributed system are always different because of the 28 physical limitations of clock synchronization [32] in distributed systems. As a result, a process might experience that the time has flown backwards, when it migrates to a node whose clock lags behind the clock of the node from which the process has migrated. Condor and Sprite have solved this problem by executing all the location dependent calls ....

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice Hall, 1995.

No context found.

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice-Hall, Englewood Cliffs, NJ 07632, USA, 1995.

No context found.

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice Hall, 1995.

No context found.

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice-Hall, Englewood Cliffs, NJ, USA, 1995.

No context found.

Andrew S. Tanenbaum. Distributed Operating Systems, chapter 6, pages 289--375. Prentice-Hall, Inc., 1995.

No context found.

Andrew S. Tanenbaum, Distributed Operating Systems, Prentice Hall, 1995

No context found.

Andrew S. Tanenbaum. Distributed Operating Systems. Prentice Hall, 1995.

First 50 documents  Next 50

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