S. Zhou, J. Wang, X. Zheng, and P. Delisle, "Utopia: a load-sharing facility for large heterogeneous distributed computing systems", Software: Practice and Experience, Vol. 23, No. 2, 1993, pp. 1305-1336.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of target bandwidth and window size. 6. Related work Previous work on exploiting available idle time on workstation clusters used a conservative model that would only run processes when the local user was away from their workstation, and no local processes were runnable. Condor [12] LSF [22], and NOW [2] use variations on a social contract to strictly limit interference with local users. However, even with these policies, there is some disruption of the local user when they return since the guest process must be evicted and the local state restored. The Linger Longer approach ....

S. Zhou, X. Zheng, J. Wang, and P. Delisle, "Utopia: a Load Sharing Facility for Large, Heterogeneous Distributed Computer Systems," SPE, 23(12), 1993, pp. 1305-1336.

....with the highest priority or longest waiting time; it will try to start new jobs on the most suitable or least loaded queue. The Load Sharing Facility (LSF) is a commercial system from Platform Computing Corp. PLATFORM] LSF evolved from the Utopia system developed at the University of Toronto [Zhou93] and is currently the most widely used commercial job management system. LSF comprises distributed load sharing and batch queuing software that manages, monitors and analyses the resources and workloads on a network of heterogeneous computers, and has fault tolerance capabilities. 3.4.2 Storage ....

S. Zhou, X. Zheng, J. Wang, and P. Delisle, "Utopia: A load sharing facility for large heterogeneous distributed computer systems." Software Practice and Experience, December 1993.

....are capable of migrating active jobs, most don t have implicit load balancing policies implemented. Most systems only allow for explicit load balancing. That is, there is no load balancing policy; the user decides which processes to migrate, and when. For example, Accent [93] Locus [76] Utopia [95], DEMOS MP [59] V [75] NEST [1] and MIST [14] fall into this category. A few systems do have implicit load balancing policies, however they are strictly non preemptive policies (active processes are only migrated for purposes other than load balancing, such as preserving workstation autonomy) ....

....[6] Sprite [20] Condor [52] and Mach [54] fall into this category. In general, non preemptive load balancing strategies depend on a priori information about processes; e.g. explicit knowledge about the runtimes of processes or user provided lists of migratable processes ( 1] 51] 20] [95]) Only a few systems have implemented automated preemptive load balancing poli 73 cies: MOSIX [7] and RHODOS [27] Preemptive migration has been found to be very effective for load balancing in MOSIX, 7] 8.2.2 Studies Migration policies have been studied much more in simulations and ....

[Article contains additional citation context not shown here]

S. Zhou, J. Wang, X. Zheng, and P. Delisle. Utopia: a load-sharing facility for large heterogeneous distributed computing systems. Software -- Practice and Expeience, 23(2):1305--1336, December 1993.

....network of workstations (NOW) that is available for serving large (parallel) jobs. Such a system requires an effective policy for recruiting idle nodes as well as efficient mechanisms for migrating the processes of parallel jobs away from nodes that are preempted by a higher priority user [2, 30, 1]. Although we do not consider the impact of node interruptions nor particular policy customizations that might be needed, we consider synthetic workloads and repartitioning overheads that are relevant to such environments. Repartitioning overheads are discussed further in section 4. 3.2 Synthetic ....

S. Zhou, J. Wang, X. Zheng, P. Delisle, Utopia: A Load Sharing Facility for Large Heterogeneous Distributed Computing Systems. Technical Report, University of Toronto, 1992.

....into two categories: solutions requiring a new operating system, and solutions requiring only network daemons. We present a brief overview of these two techniques and cite several implementations. We refer the reader to [8] for a discussion of the features of distributed operating systems, and to [9] for the details of process distribution systems. 2.1 The Operating System Approach In distributed operating systems an entire process is moved to a remote processor and restarted. Moving a process requires that both its code and data be migrated to the new processor. In the best case, process ....

....The process is not moved to the other machine by the process distribution scheme; these systems typically require a uniform file system so that all programs are accessible by all machines in the network. Prototype process distribution systems of this type include Condor [1] Butler [5] Utopia [9], and the Portable Batch System [16] Commercial systems include TaskBroker by Hewlett Packard [18] and Jobware by Ockham Technology [19] While these systems and less complex than the operating system approach, they do require the installation of network daemons on each machine participating in ....

[Article contains additional citation context not shown here]

S. Zhou, J. Wang, X. Zheng, P. Delisle, "UTOPIA: a load sharing facility for large, heterogeneous distributed computer systems," Software - Practice and Experience, 23(12):1305-- 1336, December 1993.

....primary replica is detected, the time to identify and advertise the identity of the former backup replica that is now the primary replica is approximately 3.7 milliseconds. 10. Related Work Over the past decade, a number of resource management systems for cluster computing have been implemented [12, 21, 23, 25]. A survey of 20 research and commercial cluster management systems can be found in [1] Excluding the management fault tolerance features, the basic functionality of the FTCT is currently similar to the functionality of the GLUnix system [12] While various projects mention the possibility of ....

S. Zhou, X. Zheng, J. Wang, and P. Delisle, "Utopia: A Load Sharing Facility for Large, Heterogeneous Distributed Computer Systems," Software - Practice and Experience 23(12 ), pp. 1305-1336 (December 1993).

....systems due to the increasing costs of distributed search. Distributed 35 clustering takes a middle ground by performing centralized load sharing within each resource cluster and distributed load sharing between clusters, resulting in improved scalability. The Utopia load sharing facility [73, 81] uses a cluster architecture for scalable distribution of resource load information. Load information can be exchanged between clusters according to a configured directed graph. The default job placement policy prefers to schedule jobs in the local cluster but jobs will run in remote clusters if a ....

S. Zhou, J. Wang, X. Zheng, and P. Delisle. Utopia: A load-sharing facility for large heterogeneous distributed computing systems. Software -- Practice and Experience, 23(2):1305--1336,

....during which we have demonstrated Legion features numerous times, conducted tutorials on multiple occasions and supported various academic users running a variety of applications. A. Queuing Systems Traditionally, queue systems have been used to schedule jobs on a cluster of nodes [2] 7] 11] [12]. When a user submits a job, the queue provides her with a ticket or job ID or token, which can be used to monitor the job at any later time. The ticket becomes invalid shortly after the job completes. Most queuing systems comply with a POSIX interface that requires three standard tools for ....

Zhou, S., Wang, J., Zheng, X., Delisle, P., "Utopia: A Load Sharing Facility for Large, Heterogeneous Distributed Computer Systems", Software Practice and Experience, Vol. 23, No. 2, 1993.

....months during which we have demonstrated Legion features numerous times, conducted tutorials on multiple occasions and supported various academic users running a variety of applications. 3.1. Queuing Systems Queuing systems have been used to schedule jobs on many clusters of nodes [2] 7] 11] [12]. When a user submits a job, the queue provides a ticket or job ID or token, which can be used to monitor the job at any later time. The ticket becomes invalid shortly after the job completes. Most queuing systems comply with a POSIX interface requiring three standard tools for running jobs: a ....

Zhou, S., Wang, J., Zheng, X., Delisle, P., "Utopia: A Load Sharing Facility for Large, Heterogeneous Distributed Computer Systems", Software Practice and Experience, Vol. 23, No. 2, 1993.

....primary replica is detected, the time to identify and advertise the identity of the former backup replica that is now the primary replica is approximately 3.7 milliseconds. 6. Related Work Over the past decade, a number of resource management systems for cluster computing have been implemented [10, 19, 21, 23]. A survey of 20 research and commercial cluster management systems can be found in [2] Excluding the management fault tolerance features, the basic functionality of the FTCT is currently similar to the functionality of the GLUnix system [10] While various projects mention the possibility of ....

S. Zhou, X. Zheng, J. Wang, and P. Delisle, "Utopia: A Load Sharing Facility for Large, Heterogeneous Distributed Computer Systems," Software - Practice and Experience 23(12 ), pp. 1305-1336 (December 1993).

....distributed operating systems, scheduling in distributed systems, toolkits for distributed computing, and parallel processing in distributed systems. 2.2. 1 Distributed Operating Systems An active area of research in distributed operating systems is the accommodation of heterogeneity [8] 67] 68][98]. Many of these systems deal with heterogeneity of many kinds including processor type and file system differences. Much of this research is concerned with accommodating these heterogeneities in a transparent manner. Few of these systems attempt to exploit heterogeneity since high performance is ....

....format differences, alignment differences, byte ordering differences, and size differences. The differences may be due to the hardware, operating system, or the compilers used. If formats differ in the range of values that can be represented, it may not be possible to perform a transformation [98]. Data format conversion is handled in one of two ways, either a common format such as XDR [82] is used or application specific conversions are employed. A common format requires both encoding and decoding of data while the application specific conversions are one way only and are much less ....

[Article contains additional citation context not shown here]

S. Zhou et al, "Utopia: A Load Sharing Facility for Large Heterogeneous Distributed Computer Systems," Software: Practice and Experience, Vol. 23(12), December 1993.

....more of a load on their host processors. 5. RELATED WORK Previous work on exploiting available idle time on workstation clusters used a conservative model that would only run processes when the local user was away from their workstation, and no local processes were runnable. Condor [11] LSF [24], and NOW [3] use variations on a social contract to strictly limit interference with local users. However, even with these policies, there is some disruption of the local user when they return since the guest process must be evicted and the local state restored. The LingerLonger approach ....

S. Zhou, X. Zheng, J. Wang, and P. Delisle, "Utopia: a Load Sharing Facility for Large, Heterogeneous Distributed Computer Systems," SPE, 23(12), 1993, pp. 1305-1336.

....this is inapplicable when receiving the initial request from a remote user process. Another approach is to use a challenge based on the file system security mechanisms, whereby the daemon challenges the requesting process to read the contents of a file that is accessible only to the claimed user [44]. The major drawback in this solution is that it is NFS based, and limits the usage of the system to NFS based configurations. A better alternative is using a secured authentication protocol such as MIT Kerberos. Windows 2000 and also some recent Linux distributions implement the Kerberos v5 ....

S. Zhou, X. Zheng, J. Wang, and P. Delisle, "Utopia: a load sharing facility for large, heterogeneous distributed computer systems ". Software --- Pract. & Exp. 23(12), pp. 1305--1336, Dec 1993. 28

No context found.

S. Zhou, J. Wang, X. Zheng, and P. Delisle, "Utopia: a load-sharing facility for large heterogeneous distributed computing systems", Software: Practice and Experience, Vol. 23, No. 2, 1993, pp. 1305-1336.

No context found.

S. Zhou, J. Wang, X. Zheng, and P. Delisle, "UTOPIA: A load sharing facility for large, heterogeneous distributed computing systems," Tech. Rep. CSRI-257, Computer Systems Research Institute, University of Toronto, April 1992.

No context found.

Zhou, S., et al., "Utopia: A Load-Sharing Facility for Large, Heterogeneous Distributed Computer Systems," Software: Practice and Experience, Vol. 23, No. 12, Dec. 1993, pp. 1305--1336.

No context found.

S. Zhou, J. Wang, X. Zheng, and P. Delisle. Utopia: A load-sharing facility for large heterogeneous distributed computing systems. Software -- Practice and Experience, 23(12), 1993.

No context found.

S. Zhou, J. Wang, X. Zheng, and P. Delisle. Utopia: A load-sharing facility for large heterogeneous distributed computing systems. Software -- Practice and Experience, 23(12), 1993.

No context found.

S. Zhou, X. Zheng, J. Wang, and P. Delisle, "Utopia: a load sharing facility for large, heterogeneous distributed computer systems ". Software --- Pract. & Exp. 23(12), pp. 1305--1336, Dec 1993.

No context found.

Zhou S., Zheng X., Wang J., Delisle P., Utopia: A Load Sharing Facility for Large Heterogeneous Distributed Computer Systems, Software - Practice and Experience, 23(12), 1993, pp1305-1336.

No context found.

S. Zhou, X. Zheng, J. Wang, and P. Delisle. "Utopia: a load sharing facility for large, heterogeneous distributed computer systems". Software: Practice & Experience, 23(12):1305-1336, December 1993.

No context found.

Zhou, S., et al., "Utopia: A Load-Sharing Facility for Large, Heterogeneous Distributed Computer Systems," Software: Practice and Experience, Vol. 23, No. 12, Dec. 1993, pp. 1305--1336.

No context found.

Zhou, S., Zheng, X., Wang, J. and Delisle, P., "Utopia: A load sharing facility for large, heterogeneous distributed computer systems", Software-Practice and Experience, 23(12), pp 13051336, December 1993.

No context found.

S. Zhou, J. Wang, X. Zheng, and P. Delisle, "Utopia: A Load-sharing Facility for Large Heterogeneous Distributed Computing Systems," Software - Practice and Experience, vol. 23, no. 2, pp. 1305-1336, December, 1993.

No context found.

S. Zhou, J. Wang, X. Zheng, and P. Delisle, "Utopia: A Load Sharing Facility for Large, Heterogeneous Distributed Computer Systems," SPE, vol. 23, pp. 1305-1336, 1993.

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