Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Thesis.pdf (5.06 MB)

ETD Abstract Container

Abstract Header

Resource Allocation using Adaptive Characterization of Online, Data-Intensive Workloads

Kelley, Jaimie
http://orcid.org/0000-0003-0649-8383
http://rave.ohiolink.edu/etdc/view?acc_num=osu1499949901693382

Abstract Details

2017, Doctor of Philosophy, Ohio State University, Computer Science and Engineering.
Cloud resource providers balance maximizing utilization under a power cap with meeting workload Service Level Agreements (SLA). As the amount of data used by workloads increases, so do the pressures on compute capacity in the cloud. Even if the resources assigned meet an interactive workload’s need for low latency, the data that interactive workload processes with allocated resources may not be sufficient to achieve a standard of answer quality. Increasing the resources allocated to a specific workload to meet its answer quality standard reduces the overall profit a cloud provider can make on interactive workloads. However, if a workload’s answer quality standard is not met, the interactive workload may seek another placement. Cloud instances can be purchased by the minute, and multiple opportunities for placement exist. Because of this, cloud providers need to put their clients’ interests first or lose revenue. To best serve their own and their clients interest, cloud providers need data which reflects resource usage, answer quality, and service level. If a cloud provider knows the amount of power used by each workload scheduled, it can better fulfill its power cap requirements without penalty. If a cloud provider knows the current latency and answer quality of scheduled workloads, it can decide when to reallocate resources. However, this is difficult because any collection of data online imposes overheads. While cloud providers generally reserve some percentage (5%) of utilization for operating system functions, data collection and analysis must be done carefully to avoid undue impact on scheduled workloads. I use adaptive solutions to trade accuracy for overhead in workload characterization. Adaptive workload characterizations inform resource management without the high overhead of complete calculation, but are not completely accurate. In my work, I adaptively reduce the time spent profiling peak power to the degree of accuracy that a cloud provider is willing to accept. I developed a model for adaptively profiling peak power usage to determine core scaling. Adaptive profiling saved up to 93% collection time while reducing accuracy by 3% on average. To obtain answer quality for online resource management, I overlap execution of online requests with the execution of requests that use all relevant data by using memoization of complete responses from specific components. I built Ubora to obtain and allow management of answer quality for interactive, data-intensive workloads. Cloud providers set the rate at which queries are sampled, which exchanges overhead for accuracy. Finally, I designed Quikolo, a service that speculatively deploys and characterizes a target workload in-situ in a colocation placement. Clients use this characterization to decide whether to migrate their workload to this available placement. Quikolo also enables study of overhead and accuracy influenced by the number of features and collection time used for workload characterization. Adaptively trading accuracy reduces the impact of workload characterization on overhead. My adaptive characterization solutions enable cloud providers to provision for lower overhead and still achieve information that aids balancing client needs with available cloud resources.
Christopher Stewart, Ph.D. (Advisor)
Srinivasan Parthasarathy, Ph.D. (Committee Member)
P. Sadayappan, Ph.D. (Committee Member)
187 p.
Computer Engineering; Computer Science
workload characterization; answer quality; adaptive accuracy; core scaling

Recommended Citations

Citations

  • Kelley, J. (2017). Resource Allocation using Adaptive Characterization of Online, Data-Intensive Workloads [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1499949901693382

    APA Style (7th edition)

  • Kelley, Jaimie. Resource Allocation using Adaptive Characterization of Online, Data-Intensive Workloads. 2017. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1499949901693382.

    MLA Style (8th edition)

  • Kelley, Jaimie. "Resource Allocation using Adaptive Characterization of Online, Data-Intensive Workloads." Doctoral dissertation, Ohio State University, 2017. http://rave.ohiolink.edu/etdc/view?acc_num=osu1499949901693382

    Chicago Manual of Style (17th edition)

osu1499949901693382
1,062
© 2017, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.