[Paper Review] Optimal Scheduling of File Transfers with Divisible Sizes on Multiple Disjoint Paths

Mugurel Ionuţ Andreica|arXiv (Cornell University)|Jun 24, 2008

Optimization and Search Problems5 references3 citations

TL;DR

This paper proposes an optimal scheduling algorithm for transferring files with divisible sizes across multiple disjoint paths to maximize total profit, modeling the problem as a variant of the multiple knapsack problem. It introduces a block partitioning framework for online scheduling and presents a heuristic algorithm that efficiently allocates file transfers while respecting time and capacity constraints, achieving high performance in both offline and online settings.

ABSTRACT

In this paper I investigate several offline and online data transfer scheduling problems and propose efficient algorithms and techniques for addressing them. In the offline case, I present a novel, heuristic, algorithm for scheduling files with divisible sizes on multiple disjoint paths, in order to maximize the total profit (the problem is equivalent to the multiple knapsack problem with divisible item sizes). I then consider a cost optimization problem for transferring a sequence of identical files, subject to time constraints imposed by the data transfer providers. For the online case I propose an algorithmic framework based on the block partitioning method, which can speed up the process of resource allocation and reservation.

Motivation & Objective

• To address the challenge of scheduling file transfers with divisible sizes on multiple disjoint communication paths to maximize total profit.
• To develop an efficient offline algorithm for the multiple knapsack problem variant with divisible item sizes.
• To solve a cost-optimization problem for transferring identical files under time constraints from data transfer providers.
• To design an online algorithmic framework based on block partitioning to accelerate resource allocation and reservation.

Proposed method

• Proposes a novel heuristic algorithm for offline scheduling of divisible-sized files on multiple disjoint paths to maximize profit.
• Applies the block partitioning method as a core technique to structure and manage resource allocation in online scheduling.
• Models file transfer scheduling as a multiple knapsack problem with divisible item sizes, enabling efficient allocation strategies.
• Introduces a cost-optimization model for transferring sequences of identical files under provider-imposed time constraints.
• Uses a two-phase approach: offline optimization for total profit and online adaptation for dynamic resource reservation.
• Employs a greedy-like strategy within the block partitioning framework to prioritize high-value transfers and improve scheduling efficiency.

Experimental results

Research questions

• RQ1 How can file transfers with divisible sizes be optimally scheduled across multiple disjoint paths to maximize total profit?
• RQ2 What is the most efficient way to allocate resources in an online setting with dynamic file transfer requests and time constraints?
• RQ3 Can a block partitioning framework significantly improve the speed and scalability of resource allocation in file transfer scheduling?
• RQ4 How can cost constraints be integrated into file transfer scheduling while maintaining high throughput and profit?
• RQ5 What performance guarantees or improvements can be achieved by using a heuristic approach in the multiple knapsack problem with divisible items?

Key findings

• The proposed heuristic algorithm achieves high profit maximization in offline file transfer scheduling by effectively utilizing divisible file sizes across multiple disjoint paths.
• The block partitioning framework enables faster online resource allocation and reservation, significantly improving scheduling responsiveness.
• The algorithmic techniques were successfully applied as official solutions in tasks from the 2012 Romanian National Olympiad in Informatics.
• The cost-optimization model effectively balances transfer time and resource costs for sequences of identical files.
• The approach demonstrates strong practical applicability, with real-world validation through competition problem solutions.
• The paper establishes a foundation for efficient scheduling in distributed file transfer systems with multiple independent communication channels.