A survey of load sharing in networks of workstations

Guy, Bernard; Stève, Dominique; Simatic, Michel

doi:10.1088/0967-1846/1/2/003

Cited by 15 publications

(14 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A comparison of CENTRAL and an equivalent distributed algorithm [The89] indicated that the simplicity of implementation of the latter can be an advantage. Other work has highlighted further potential weaknesses of centralised algorithms, notably bottlenecks forming at the central node and the vulnerability of the systems load sharing capabilities if this node fails [Ald92,Ber93]. These factors have lead to the conclusion that centralised solutions are better suited to multi-processor configurations, rather than distributed systems.…”

Section: Survey Of Related Researchmentioning

confidence: 99%

“…The trend now is to use three: transfer, information and location [Zho88,Gha90,Bak92,Ber93,Mah93,Kru94,Ben95].…”

Section: Survey Of Related Researchmentioning

confidence: 99%

“…Although very simple, this form of "blind" [Ber93] location policy can exhibit substantial performance improvement over the no load sharing case at all levels of system load when implemented on homogeneous systems. Performance on heterogeneous systems is discussed in later chapters of this thesis.…”

Section: Location Policymentioning

confidence: 99%

See 2 more Smart Citations

Evaluation of loadsharing algorithms for heterogeneous distributed systems

Leslie

McKenzie

1999

Computer Communications

View full text Add to dashboard Cite

Distributed systems offer the ability to execute a job at other nodes than the originating one. Load sharing algorithms use this ability to distribute work around the system in order to achieve greater efficiency. This is reflected in substantially reduced response times. In the majority of studies the systems on which load sharing has been evaluated have been homogeneous in nature. This thesis considers load sharing in heterogeneous systems, in which the heterogeneity is exhibited in the processing power of the constituent nodes. To validate the assumptions in the simulation model a load sharing scenario was implemented on a network of Sun workstations at the University. While the scope of the implementation was somewhat limited by lack of resources, it does demonstrate the relative ease with which the algorithms can be implemented without alteration of the operating system code or modification at the kernel level.

show abstract

Section: Survey Of Related Researchmentioning

confidence: 99%

“…The trend now is to use three: transfer, information and location [Zho88,Gha90,Bak92,Ber93,Mah93,Kru94,Ben95].…”

Section: Survey Of Related Researchmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation of loadsharing algorithms for heterogeneous distributed systems

Leslie

McKenzie

1999

Computer Communications

View full text Add to dashboard Cite

show abstract

“…There have been several previous projects examining the use of remote idle memory as backing store for nodes in networks of workstations [4], [3], [5], [6], [7], [8], [9], [10], [11]. Some incorporate Network RAM into an OS's paging system, some into an OS's swapping system, and others into an OS's file system as a cooperative cache for file data.…”

Section: Related Workmentioning

confidence: 99%

Incorporating Network RAM and Flash into Fast Backing Store for Clusters

Newhall

Woos

2011

2011 IEEE International Conference on Cluster Computing

View full text Add to dashboard Cite

Abstract-We present Nswap2L, a fast backing storage system for general purpose clusters. Nswap2L implements a single device interface on top of multiple heterogeneous physical storage devices, particularly targeting fast random access devices such as Network RAM and flash SSDs. A key design feature of Nswap2L is the separation of the interface from the underlying physical storage; data that are read and written to our "device" are managed by our underlying system and may be stored in local RAM, remote RAM, flash, local disk or any other cluster-wide storage. Nswap2L chooses which physical device will store data based on cluster resource usage and the characteristics of various storage media. In addition, it migrates data from one physical device to another in response to changes in capacity and to take advantage of the strengths of different types of physical media, such as fast writes over the network and fast reads from flash. Performance results of our prototype implementation of Nswap2L added as a swap device on a 12 node Linux cluster show speed-ups of over 30 times versus swapping to disk and over 1.7 times versus swapping to flash. In addition, we show that for parallel benchmarks, Nswap2L using Network RAM and a flash device that is slower than Network RAM can perform better than Network RAM alone.

show abstract

“…There are several projects that examine using remote idle memory as backing store in clusters or networks of workstations [20], [16], [6], [2], [13], [14], [12], [8], [10]. Reliability issues are addressed in three of these projects.…”

Section: Related Workmentioning

confidence: 99%

Reliable adaptable Network RAM

Newhall

Amato

Pshenichkin

2008

2008 IEEE International Conference on Cluster Computing

View full text Add to dashboard Cite

Abstract-We present reliability solutions for adaptable Network RAM systems running on general-purpose clusters. Network RAM allows nodes with over-committed memory to swap pages over the network, storing them in the idle RAM of other nodes and avoiding swapping to slow, local disk. An adaptable Network RAM system adjusts the amount of RAM currently available for storing remotely swapped pages in response to changes in nodes' local RAM usage. It is important that Network RAM systems provide reliability for remotely swapped page data. Without reliability, a single node failure can result in failure of unrelated processes running on other nodes by losing their remotely swapped pages. Adaptable Network RAM systems pose extra difficulties in providing reliability because each node's capacity for storing remotely swapped pages changes over time, and because pages may move from node to node in response to these changes. Our novel dynamic RAID-based reliability solutions use idle RAM for storing page and reliability data, avoiding using slow disk for reliability. They are designed to work with the adaptive nature of our Network RAM system (Nswap), allowing page and reliability data to migrate from node to node and allowing pages to be added to or removed from different parity groups. Additionally, page recovery runs concurrently with cluster applications, so that cluster applications do not have to wait until all data from a failed node is recovered before resuming execution. We present results comparing Nswap to disk swapping for a set of benchmarks running on our gigabit cluster. Our results show that reliable Nswap is up to 32 times faster than swapping to disk, and that there is virtually no impact on the performance of applications as they run concurrently with page recovery.

show abstract

A survey of load sharing in networks of workstations

Cited by 15 publications

References 23 publications

Evaluation of loadsharing algorithms for heterogeneous distributed systems

Evaluation of loadsharing algorithms for heterogeneous distributed systems

Incorporating Network RAM and Flash into Fast Backing Store for Clusters

Reliable adaptable Network RAM

Contact Info

Product

Resources

About