This dissertation defines the domain of urgent computing as the set of deadline-constrained scientific simulations and models that are used to affect some critical decision (e.g., severe weather forecasting, the modeling of infectious diseases, wildfire simulations, patient-specific medical treatment). These computations typically target large, distributed systems (e.g., supercomputers, clusters, grids, clouds) that are shared by many users. This resource contention often leads to nontrivial and highly variable allocation delays that impede the ability of the computations to meet their deadlines. Thus, resources that support urgent computing should provide elevated priority access to these urgent computations. This dissertation focuses on the additional infrastructure necessary to aid these critical computations in meeting their deadlines within the current distributed computing environment. In particular, the combination of management mechanisms, access policies and tools needed to support urgent computing on batch queue resources (e.g., supercomputers and clusters) and computational clouds.
This dissertation describes the implementation of the Special PRiority and Urgent Computing Environment (SPRUCE), a token-based framework that manages urgent computing users, sessions, and resources. This framework has been widely deployed on the TeraGrid. Additionally, this dissertation proposes and evaluates a set of elevated priority policies that different resource types (i.e., batch queue resources, cloud resources) can use to support urgent computing. Included in this evaluation is a discussion of the effect these policies have on nonurgent users of the resource, along with potential compensation schemes for affected users. This dissertation also identifies and evaluates a set of statistical methods and heuristics that can be used to predict probabilistic upper bounds on the total turnaround time of the urgent computation (i.e., file staging, resource allocation, and execution delay). These upper bounds can guide the user in selecting the resource and policy that offers the greatest probability of meeting the given deadline.
|Commitee:||Beckman, Pete, Stevens, Rick|
|School:||The University of Chicago|
|School Location:||United States -- Illinois|
|Source:||DAI-B 72/12, Dissertation Abstracts International|
|Keywords:||Cloud computing, Deadline-constrained computing, Distributed computing, Elevated priority, Spruce, Urgent computing|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be