We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.DC

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Distributed, Parallel, and Cluster Computing

Title: Scheduling Parallel-Task Jobs Subject to Packing and Placement Constraints

Abstract: Motivated by modern parallel computing applications, we consider the problem of scheduling parallel-task jobs with heterogeneous resource requirements in a cluster of machines. Each job consists of a set of tasks that can be processed in parallel, however, the job is considered completed only when all its tasks finish their processing, which we refer to as "synchronization" constraint. Further, assignment of tasks to machines is subject to "placement" constraints, i.e., each task can be processed only on a subset of machines, and processing times can also be machine dependent. Once a task is scheduled on a machine, it requires a certain amount of resource from that machine for the duration of its processing. A machine can process ("pack") multiple tasks at the same time, however the cumulative resource requirement of the tasks should not exceed the machine's capacity.
Our objective is to minimize the weighted average of the jobs' completion times. The problem, subject to synchronization, packing and placement constraints, is NP-hard, and prior theoretical results only concern much simpler models. For the case that migration of tasks among the placement-feasible machines is allowed, we propose a preemptive algorithm with an approximation ratio of $(6+\epsilon)$. In the special case that only one machine can process each task, we design an algorithm with improved approximation ratio of $4$. Finally, in the case that migrations (and preemptions) are not allowed, we design an algorithm with an approximation ratio of $24$. Our algorithms use a combination of linear program relaxation and greedy packing techniques. We present extensive simulation results, using a real traffic trace, that demonstrate that our algorithms yield significant gains over the prior approaches.
Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Discrete Mathematics (cs.DM); Data Structures and Algorithms (cs.DS); Performance (cs.PF)
Cite as: arXiv:2004.00518 [cs.DC]
  (or arXiv:2004.00518v2 [cs.DC] for this version)

Submission history

From: Mehrnoosh Shafiee [view email]
[v1] Wed, 1 Apr 2020 15:42:30 GMT (2176kb,D)
[v2] Thu, 2 Apr 2020 01:34:46 GMT (2176kb,D)

Link back to: arXiv, form interface, contact.