2006
DOI: 10.1109/tpds.2006.10
|View full text |Cite
|
Sign up to set email alerts
|

Effectively utilizing global cluster memory for large data-intensive parallel programs

Abstract: Abstract-Large scientific parallel applications demand large amounts of memory space. Current parallel computing platforms schedule jobs without fully knowing their memory requirements. This leads to uneven memory allocation in which some nodes are overloaded. This, in turn, leads to disk paging, which is extremely expensive in the context of scientific parallel computing. To solve this problem, we propose a new peer-to-peer solution called Parallel Network RAM. This approach avoids the use of disk, better uti… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
2
0
2

Year Published

2009
2009
2015
2015

Publication Types

Select...
5
1

Relationship

0
6

Authors

Journals

citations
Cited by 16 publications
(4 citation statements)
references
References 36 publications
0
2
0
2
Order By: Relevance
“…As a scope of future research, the time-efficiency of ParSym algorithm may be improved further using the bounded-collision parallel memory-mapping scheme for effective utilization of cluster memory [7,21]. Moreover, incorporation of spatial information in the feature vector as this is found to be effective in pixel classification [1] in ParSym method, in lieu of intensity values at different spectral bands constitutes an important direction for farther research.…”
Section: Resultsmentioning
confidence: 99%
“…As a scope of future research, the time-efficiency of ParSym algorithm may be improved further using the bounded-collision parallel memory-mapping scheme for effective utilization of cluster memory [7,21]. Moreover, incorporation of spatial information in the feature vector as this is found to be effective in pixel classification [1] in ParSym method, in lieu of intensity values at different spectral bands constitutes an important direction for farther research.…”
Section: Resultsmentioning
confidence: 99%
“…Among the numerous technologies in HPC, we note that grid computing [7], [8], [9] and cluster [10], [11], [12] are rapidly progressing. Cluster as a unified computing resource is a parallel or distributed system and is made up of many connected and similar elements.…”
Section: Introductionmentioning
confidence: 99%
“…O desenvolvimento de técnicas e métricas eficientes para distribuir processos a nós do sistemaé umaárea tradicional em sistemas distribuídos, que ainda não foi totalmente explorada, sendo alvo de muitas pesquisas [13]. Um dos fatores responsáveis pela atenção dada a essa atividade em plataformas distribuídasé a relação custo/benefício.…”
Section: Considerações Iniciaisunclassified
“…No artigo [31], os autores migram a melhor tarefa, de acordo com a política de seleção utilizada, executando em um nó congestionado, para a máquina remota com a maior quantidade de memória livre. Em [13],é proposta uma nova solução ponto-a-ponto, denominada Parallel Network RAM, que procura diminuir a sobrecarga causada pela sincronização e comunicação das aplicações paralelas, evitando o uso do disco rígido, e otimizando a utilização da memória DRAM, demonstrando que o uso de esquemas diferenciadosé apropriado para situações diferentes.…”
Section: Escalonamento De Aplicações Memory-intensiveunclassified