B. Walker, G. Popek, R. English, C. Kline, and G. Thiel. The LOCUS distributed operating system. In Proceedings of the Ninth ACM Symposium on Operating Systems Principles, 1983.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....To improve the latency associated with the O(logN) lookup cost, blocks are cached on nodes that are on the path to the node that stores the primary copy. PAST [10] was designed to be a general purpose replicated object store. Systems such as Coda [7] Echo [2] Ficus [9] JetFile [6] and Locus [11] bridge between traditional distributed file systems and peer to peer storage systems. They use techniques from both camps. Many of these systems still rely on tightly coupled nodes and or partitioning the file system to achieve the desired robustness and scalability. 7 Conclusions Mammoth is a ....

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The LOCUS distributed operating system. In Proceedings of the 9th Symposium on Operating Systems Principles, Operating Systems Review, pages 49--69, October 1983.

....[2] Systems which use the Sun NFS [31] protocol use caching of file blocks in local memory to improve performance. The Cedar file system [39] introduced caching of whole files. The Andrew file system [36] provides location transparency for files and adds read only file replication. The LOCUS [45] provides consistent sharing with a sophisticated locking and transactions mechanism for shared files. It also supports sharing of such files. This results in a complex interface and implementation. There have been some studies of the performance of existing file systems. Satyanarayanan ....

....normal operation. Client caches are kept for a maximum of 30 seconds as the NFS protocol specifies. In the case of disk failures, and the other recovers, inconsistent data will be stored. HA NFS do not detect the last process to fail and can then use an old version for recovery. Locus In LOCUS [45] is a system with a single name space for all objects. It is not possible to detect the location of an object by using its name, they use pure names [22] It is possible to use different degrees of replication. A file belongs to a file group. The file group has a set of sites with physical ....

Walker, B., Popek, G., English, R., Kline, C., and Thiel, G. The LOCUS distributed operating system. In Proceedings of the 9th Symposium on Operating System Principles (October 1983), pp. 49--70.

....case of device failures. We plan to study the possibility of using SC in the iSCSI protocol. 9 Related Work SC is closely related to work in process migration, faulttolerant operating systems, and in providing high availability for Internet services through protocol support. Process migration [14, 30, 4, 9] is a heavy weight, generic mechanism that enables seamless execution of an application process at multiple nodes during its lifetime, targeting load sharing and balancing in clusters. SC differs from classical process migration in that it trades off transparency for finer migration granularity, ....

....not freeze execution, and does not require inter process or client server synchronization. This comes at the expense of potential re execution after a migration. Most existing systems supporting process migration are custom operating systems designed from scratch with builtin migration support [30, 4, 9]. In contrast, we provide a fairly straightforward implementation of SC in a generalpurpose OS. The idea of using logs at the OS level for deterministic execution replay can be traced back to early fault tolerant operating systems like NonStop [5] and Auros [6, 7] In the NonStop kernel, a ....

B. Walker, G. Popek, R. English, C. Kline, and G. Thiel. The LOCUS Distributed Operating System. In Proc. 9th Symp. on Operating Systems Principles (SOSP), pages 49-- 70, 1983.

....applications. Moreover, applications must be involved n the detection and resolution of conflicts since these naturally depend on the semantics of the application. To this end, Bayou provides system support for applicationspecific conflict detection and resolution. Previous systems, such as Locus [30] and Coda [17] have proven the value of semantic conflict detection and resolution for file directories, and several systems are exploring conflict resolution for file and database contents [8, 18, 26] Bayou s mechamsms extend this work by letting applications exploit domain specific knowledge ....

....the procedures used to handle conflicts, and, more generally, in their ability to deal with conflicts. Techniques for semantic based conflict detection and resolution have previously been incorporated into some systems to handle special cases such as file directory updates. For example, the Locus [30], Flcus [12] and Coda [17] distributed file systems all include mechanisms for automatically resolving certain classes of conflicting directory operations. More recently, some of these systems have also incorporated support for resolver programs that reduce the need for human intervention when ....

B. Walker, G. Popek, R. English, C. Kline, and G. Thiel. The LOCUS distributed operating system. Proceedings Ninth Symposium on Operating Systems Principles, Bretton Woods, New Hampshire, October 1983, pages 49-70. 183

....have we violated this principle. Another scalability principle we have adopted is the avoidance of system wide rapid change. Consequently, we have rejected strategies that require election or agreement by large numbers of nodes. For example, we have avoided algorithms such as that used in Locus [23] that depend on nodes achieving consensus on the current partition state of the network. 3.2 Portable Workstations Powerful, lightweight and compact laptop computers are commonplace today. It is instructive to observe how a person with data in a shared file system uses such a machine. ....

....updated at the client or the server has been deleted by the other, or if directory attributes have been modified at the server and the client. This strategy of resolving partitioned directory updates is consistent with our strategy in server replication [11] and was originally suggested by Locus [23]. Our original design for disconnected operation called for preservation of replay files at servers rather than clients. This approach would also allow damage to be confined by marking conflicting replicas inconsistent and forcing manual repair, as is currently done in the case of server ....

[Article contains additional citation context not shown here]

WALKER, B., POPEK, G., ENGLISH, R., KLINE, C., AND THIEL, G. The LOCUS distributed operating system. In Proceedings of the 9th ACM Symposium on Operating System Principles (Bretton Woods, N.H., Oct. 1983), pp. 49-70. Received June 1991; revised August 1991; accepted September 1991 ACM Transactions on Computer Systems, Vol. 10, No. 1, February 1992.

....its operations (e.g. rename ) affect multiple files, each replicated on a different set of nodes. Such operations demand a new protocol for ensuring consistent outcome after conflicts, as we discuss in Section 5.2. Pangaea offers a simple conflict resolution policy similar to that of Roam, Locus [36], or Coda [18] We chose this design over more sophisticated approaches (as in Bayou) because Pangaea can make no assumptions about the semantics of file system operations. FARSITE [1] and Pangaea both build a unified file system across a federation of nodes, but they have different objectives. ....

....this case, the clocks of servers should be loosely synchronized, e.g. using NTP, to respect the users intuitive sense of update ordering. The second option is to concatenate two versions in the file and let the user fix the conflict manually. Other options, such as application specific resolvers [36, 18, 32], are certainly possible, but we have not implemented them yet. Conflicts regarding file attributes or directory entries are more difficult to handle. They fall into two categories. The first is a conflict between two directory update operations; for example, Alice does mv foo alice foo and ....

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The Locus distributed operating system. In 9th Symp. on Op. Sys. Principles (SOSP), pages 49--70, Bretton Woods, NH, USA, October 1983.

....VV entry pruning. Section 1.3 overviews VV pruning algorithms proposed in the past. Section 2 presents our algorithm, and Section 3 proves its correctness. 1. 1 Description of the basic VV algorithm Figure 1 shows the basic VV algorithm, introduced first in the LOCUS distributed operating system [11, 16]. A VV is a table that partially maps a node ID to a timestamp. It is kept for each copy (replica) of an object replicated in a distributed system. A node ID is any bit string that uniquely identifies a node that stores a replica (e.g. an IP address) A timestamp is any monotonically increasing ....

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The Locus distributed operating system. In 9th Symp. on Op. Sys. Principles (SOSP), pages 49--70, Bretton Woods, NH, USA, October 1983. 8

....any replica be removed from the system and yet ensures the consistency of other replicas. LBFS [5] saves network bandwidth by exchanging fingerprints of file fragments instead of file contents. Its idea complements ours; while LBFS reduces bandwidth consumption, we reduce access latency. Locus [9] provided a file system with optimistic, fine grain replication. Pangaea extends Locus by supporting frequent replica addition and removal. Mobile data sharing services, such as Ficus, Roam [7] RefDBMS [2] and Bayou [6] allow mobile users to replicate data massively and exchange updates ....

....a graph edge is spanned only between two live replicas, and no orphan replicas, disconnected from others, are created because of merging. Concurrent updates to file contents are also detected using version vectors and merged semantically. We currently use a protocol similar to that of Locus [9]. 3.3 Maintaining the consistency of the hierarchical name space The previous sections described Pangaea s decentralized approach for maintaining links among replicas of a single file. In fact, because Pangaea offers a hierarchical name space, it must also manage links between a file and its ....

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The LOCUS distributed operating system. In 9th Symp. on Op. Sys. Principles (SOSP), pages 49--70, Bretton Woods, NH, USA, October 1983. 5

....servers using a network virtual disk function. This approach has many of the same drawbacks as xFS. 175 HAMFS deploys many of the similar techniques developed for increased performance and availability used in many of the documented file systems, such as Cedar [14] Echo [15] XFS [16] Locus [17], HARP[18] and Spritely NFS [19] Additionally, HAMFS offers features that others do not, including Token escalation, Space reserve, and automatic deadlock detection. 7 Conclusions The asymmetric shared file system organization is a superior approach for implementing a commercial cluster file ....

B. J. Walker, G. J. Popek, R. English, C. S. Kline, and G. Thiel. The locus distributed operating system. In proceedings of Ninth ACM Symposium on Operating System Principles, pages 49-70, Bretton Woods, New Hampshire, October 1983.

....Classical CBCAST implementations [11] of replication strictly rely on message orderings, without incorporating the application level information used for mirroring in our infrastructure. Our checkpointing implementation relies on past work on transactional processing in distributed systems [18, 19]. Slice uses mirroring techniques to reliably mirror files and thereby support failure atomic file operations in a distributed, scalable network storage system [20] The TACT [21] project also chooses replication as a means of increasing service availability, and it provides a middleware layer ....

B. Walker et al., "The LOCUS Distributed Operating System ", In Proc. of the Ninth ACM Symposium on Operating Systems Principles, pp. 49-70, Dec. 1983

....dynamic evaluation, the implementer need only identify how the object will be accessed. The dynamic evaluation algorithms manage concurrency and compatibility. 10 Related Research Many distributed file systems clients transact with a server on every open and close of a file for synchronization [25, 5, 26]. This is the simplest technique to implement local file system open semantics in the distributed environment. However, all file open and file close requests require a network operation. The Andrew file system [15] interacts with a server at every open and close and uses open and close as 22 ....

B. Walker, G. Popek, R. English, C. Kline, and G. Thiel. The LOCUS distributed operating system. In Proceedings of the 9th ACM Symposium on Operating Systems Principles, 1983.

....[11] the successor of Telescript, and Aglets [15] Systems such as Telescript [30] Agent Tcl [12] and ARA [19] provide strong mobility by using a dedicated language interpreter to capture and resume the process execution state. Full mobility is provided by LOCUS distributed operating system [29]. Full mobility is necessary if process migration is used, for instance, for load balancing: the migration has to be completely transparent. We would say that the notion of mobility at the heart of the classical concept of mobile agent is strong mobility: the execution state of a migrating agent ....

....its state on secondary memory, and upon resumption continues its execution) We shall experiment on this. We would like to conclude by comparing our work with other similar approaches. The idea of migrating processes was already exploited in the 1980s, in the LOCUS distributed operating system [29], for load balancing. Some other systems [9, 18] use state saving techniques to provide transparent process migration or persistence functionalities (a survey of such techniques can be found in [23] However, in these systems, the migration mechanisms are part of the operating system or of the ....

B. Walker, G. Popek, R. English, C. Kline, and G. Thiel. The LOCUS distributed operating systems. In Proceedings of the 9th ACM Symposium on Operating Systems Principles (SOSP), volume 17, pages 49-70, 1983.

....no consistency problems. Alternately, the mobile computer can hold the writeable copy, but only at the cost of making the data unwriteable by anyone else. Optimistic replication allows replica servers to access and update data records independent from one another by delaying consistency checks [26]. Instead of preventing inconsistency from occurring, mechanisms for detecting and resolving inconsistency are used. Periodically, replicas exchange updates with one another in a process called reconciliation. During the reconciliation process, two replicas exchange all updates that occurred ....

....resolution allows some form of disconnected operation, is has the obvious problem of 4 potential loss of data. For most real world situations, a more capable mechanism for conflict detection and resolution is needed. Optimistic replication can use a client server model [20] or a peer model [26]. In the client server model, one replica of the data is designated as the special server replica. All updates created at other replicas must be registered with the server before they can be propagated further. The client server model tremendously simplifies the design of a replicated system, at ....

B. Walker; G. Popek; R. English; C. Kline; G. Thiel, "The LOCUS distributed operating system," Operating Systems Review, vol.17, (no.5, spec. issue.), pp.49-70, 1983.

No context found.

B. Walker, G. Popek, R. English, C. Kline, and G. Thiel. The LOCUS distributed operating system. In Proceedings of the Ninth ACM Symposium on Operating Systems Principles, 1983.

No context found.

Bruch Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The LOCUS distributed operating system. In Proceedings of the 9th Symposium on Operating Systems Principles (SOSP), pages 49--70, November 1983.

No context found.

Walker, B., Popek, G., English, R., Kline, C., Thiel, G.: The LOCUS Distributed Operating System. In: Proc. of the 9th ACM Symposium on Operating Systems Principles. (1983)

No context found.

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The locus distributed operating system. In Proceedings of the 9th ACM Symposium on Operating Systems Principles, pages 49-70, 1983.

No context found.

WALKER, B., POPEK, G., ENGLISH, R., KLINE, C., AND THIEL, G. The locus distributed operating system. In Proceedings of the 9th ACM Symposium on Operating Systems Principles (1983), pp. 49--70.

No context found.

Bruch Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The LOCUS distributed operating system. In Proceedings of the 9th Symposium on Operating Systems Principles (SOSP), pages 49--70, November 1983.

No context found.

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The LOCUS Distributed Operating System. ACM SIGOPS, 17(5):49-70, October 1983. 73

No context found.

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The LOCUS distributed operating system. In Proceedings of the 9th ACM Symposium on Operating Systems Principles (SOSP), volume 17, pages 4970, 1983.

No context found.

Bruce Walker, Gerald Popek, Robert English, Charles Kline, and Greg Thiel. The LOCUS distributed operating system. In Proceedings of the Ninth ACM Symposium on Operating Systems Principles, pages 49--70, Bretton Woods, NH, U.S.A., October 1983. ACM SIGOPS. 228.

No context found.

B. Walker, G. Popek, R. English, C. Kline and G. Thiel, `The LOCUS distributed operating system', Proc. 9th ACM Symposium on Operating System Principles, October 1983, pp. 49--70.

No context found.

B. Walker, G. Popek, R. English, C. Kline and G. Thiel, `The Locus distributed operating system', 9th ACM Symp. on Operating System Principles, 1983, pp. 49--70.

No context found.

Bruce Walker, Gerald Popek, Robert English, Charles Kline and Greg Thiel, `The LOCUS distributed operating system', Proceedings of the Ninth ACM Symposium on Operating Systems Principles, published in Operating Systems Review 17, (5), Bretton Woods, NH, USA, October 1983, pp. 49--70. ACM SIGOPS, ACM Press.

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