An Experimental Study of Data Migration Algorithms

Anderson, Eric; Hall, J. L.; Hartline, Jason D.; Hobbs, Michael; Karlin, Anna R.; Saia, Jared; Swaminathan, Ram; Wilkes, John

doi:10.1007/3-540-44688-5_12

Cited by 49 publications

(31 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The best approximation algorithm known for minimum edge coloring [18] then yields an algorithm for data migration with unit edge processing times, whose makespan is 1.1χ + 0.8, where χ is the chromatic index of the graph. Approximation algorithms are also developed [1,10,13,14] for generalizations of the makespan minimization problem in which there are storage constraints on disks and constraints on how the data can be transferred.…”

Section: Introductionmentioning

confidence: 99%

Combinatorial Algorithms for Data Migration to Minimize Average Completion Time

Gandhi

Mestre

2007

Algorithmica

View full text Add to dashboard Cite

The data migration problem is to compute an efficient plan for moving data stored on devices in a network from one configuration to another. It is modeled by a transfer graph, where vertices represent the storage devices, and edges represent data transfers required between pairs of devices. Each vertex has a non-negative weight, and each edge has a processing time. A vertex completes when all the edges incident on it complete; the constraint is that two edges incident on the same vertex cannot be processed simultaneously. The objective is to minimize the sum of weighted completion times of all vertices. Kim (J. Algorithms 55, 42-57, 2005) gave an LP-rounding 3-approximation algorithm when edges have unit processing times. We give a more efficient primal-dual algorithm that achieves the same approximation guarantee. When edges have arbitrary processing times we give a primal-dual 5.83-approximation algorithm. We also study a variant of the open shop scheduling problem. This is a special case of the data migration problem in which the transfer graph is bipartite and the objective is to minimize the sum of completion times of edges. We present a simple algorithm that achieves an approximation ratio of √ 2 ≈ 1.414, thus improving the 1.796-approximation given by Gandhi et al. (ACM Trans. Algorithms 2(1), [116][117][118][119][120][121][122][123][124][125][126][127][128][129] 2006). We show that the analysis of our algorithm is almost tight.

show abstract

Section: Introductionmentioning

confidence: 99%

Combinatorial Algorithms for Data Migration to Minimize Average Completion Time

Gandhi

Mestre

2007

Algorithmica

View full text Add to dashboard Cite

show abstract

“…More recently, Hall et al, 2001 [14], Saia, 2001 [20] and Anderson et al, 2001 [2] have studied various flavours of a problem, referred to as the data migration problem, which consists in computing an efficient plan for moving objects stored on devices in a fully connected network from one configuration to another. On top of requiring that each device is involved in the transfer of only one object at a time, they explicitly consider capacity constraints on each of the devices and assume both that the objects have the same size and that there is at least one free space on each storage device in the initial as well as in the final configuration.…”

Section: Related Workmentioning

confidence: 99%

On a resource-constrained scheduling problem with application to distributed systems reconfiguration

Sirdey

Carlier

Kerivin

et al. 2007

European Journal of Operational Research

View full text Add to dashboard Cite

International audienceThis paper is devoted to the study of a resource-constrained scheduling problem, the Process Move Programming problem, which arises in relation to the operability of certain high availability real-time distributed systems. Informally, this problem consists, starting from an arbitrary initial distribution of processes on the processors of a distributed system, in finding the least disruptive sequence of operations (non-impacting process migrations or temporary process interruptions) at the end of which the system ends up in another predefined arbitrary state. The main constraint is that the capacity of the processors must not be exceeded during the reconfiguration. After a brief survey of the literature, we prove the NP-hardness of the problem and exhibit a few polynomial special cases. We then present a branch-and-bound algorithm for the general case along with computational results demonstrating its practical relevance

show abstract

“…The dual LP contains a variable y Su (for each set S u ) corresponding to each constraint represented by (1). The dual LP is given below.…”

Section: A Linear Programming Relaxationmentioning

confidence: 99%

“…Notice that all edges assigned to u must belong to M j for some j ≤ s 1 + s 2 . We think of u as paying 1 2 of the completion time of all edges assigned to it, plus an additional 1 2 for the fully-assigned edges, which in the worst case will be scheduled the latest,…”

Section: Theorem 3 Any Strongly Minimal Schedule Is √mentioning

confidence: 99%

“…The best approximation algorithm known for minimum edge coloring [17] then yields an algorithm for data migration with unit edge length, whose makespan is 1.1χ + 0.8, where χ is the chromatic index of the graph. Approximation algorithms are also developed [9,1,13,14] for generalizations of the makespan minimization problem in which there are storage constraints on disks and constraints on how the data can be transferred.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Combinatorial Algorithms for Data Migration to Minimize Average Completion Time

Gandhi

Mestre

2006

Approximation, Randomization, and Combinatorial Optimization. Algorithms and Techniques

View full text Add to dashboard Cite

Abstract.The data migration problem is to compute an efficient plan for moving data stored on devices in a network from one configuration to another. It is modeled by a transfer graph, where vertices represent the storage devices, and the edges represent the data transfers required between pairs of devices. Each vertex has a non-negative weight, and each edge has unit processing time. A vertex completes when all the edges incident on it complete; the constraint is that two edges incident on the same vertex cannot be processed simultaneously. The objective is to minimize the sum of weighted completion times of all vertices. Kim (Journal of Algorithms, 55:42-57, 2005 ) gave an LP-rounding 3-approximation algorithm. We give a more efficient primal-dual algorithm that achieves the same approximation guarantee, which can be extended to yield a 5.83-approximation for arbitrary processing times. We also study a variant of the open shop scheduling problem. This is a special case of the data migration problem in which the transfer graph is bipartite and the objective is to minimize the completion times of edges. We present a simple algorithm that achieves an approximation ratio of √ 2 ≈ 1.414, thus improving the 1.796-approximation given by Gandhi et al.(ACM Transaction on Algorithms, 2(1):116-129, 2006). We show that the analysis of our algorithm is almost tight.

show abstract

An Experimental Study of Data Migration Algorithms

Cited by 49 publications

References 5 publications

Combinatorial Algorithms for Data Migration to Minimize Average Completion Time

Combinatorial Algorithms for Data Migration to Minimize Average Completion Time

On a resource-constrained scheduling problem with application to distributed systems reconfiguration

Combinatorial Algorithms for Data Migration to Minimize Average Completion Time

Contact Info

Product

Resources

About