Minimal Cost Reconfiguration of Data Placement in Storage Area Network

Abstract

Video-on-Demand (VoD) services require frequent updates in file configuration on the storage subsystem, so as to keep up with the frequent changes in movie popularity. This defines a natural reconfiguration problem in which the goal is to minimize the cost of moving from one file configuration to another. The cost is incurred by file replications performed throughout the transition. The problem shows up also in production planning, preemptive scheduling with set-up costs, and dynamic placement of Web applications. We show that the reconfiguration problem is NP-hard already on very restricted instances. We then develop algorithms which achieve the optimal cost by using servers whose load capacities are increased by O(1), in particular, by factor 1 + δ for any small 0 < δ < 1 when the number of servers is fixed, and by factor of 2 + ε for arbitrary number of servers, for some ε ∈ [0, 1). To the best of our knowledge, this particular variant of the data migration problem is studied here for the first time.

Heuristics for Job Scheduling Reoptimization

“…The optimal algorithm for min Σ j C j , based on a reduction to perfect matching (introduced in Section III-B), was able to handle relatively large instances of 15 machines and 300 jobs. In our basic template for job creation, the jobs' lengths were uniformly distributed in [1,20]. The job-extension penalty of job j was uniformly distributed in [1, pj 2 ], and the transition costs were uniformly distributed in [1,5].…”

“…This framework is in use also in [20], to analyze algorithms for data placement in storage area network. Job scheduling reoptimization problems with respect to the total flow-time objective was studied in [2], were algorithms for finding an optimal scheduled using the minimal possible transition cost and algorithms for optimal utilization of a limited number of migration were described.…”

Incremental List Coloring of Graphs, Parameterized by Conservation

“…Recently, there has been strong interest in analyzing the parameterized complexity of l-local search in terms of some locality parameter l [12,17]. Our locality measure c can also be seen as "transition cost" between old and new solutions-to keep this cost small has been identified as an important target, e. g., in the reconfiguration of data placements [19].…”

Reoptimization of the Minimum Total Flow-Time Scheduling Problem

“…This framework is in use also in [20], to analyze algorithms for data placement in storage area network. Considering both the quality of the solution and the transition cost from an initial solution can also be seen as a special case of multiobjective optimization problems.…”