PocketData
/
grant-2018-NSF-CRI
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Comments and Synopsis

main
Oliver Kennedy 2019-01-07 19:56:43 -05:00
parent 654e1e14f7
commit 968f6a723a
3 changed files with 6 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
Preproposal/comments.txt Normal file
View File

@ -0,0 +1 @@
The project team unites experts with a complementary skill set: (1) PI Kennedy will contribute expertise in databases, query compilers, and data structures. (2) PI Ziarek will contribute experience in compilers, program analysis, and embedded devices. (3) PI Kul will contribute expertise in data engineering, workload and behavior modeling, as well as extensive experience working with the \PocketData dataset. We have budgeted for a postdoc at UB, as well as two graduate students, one at UB, and one at DSU. Our budget also supports effort for all three investigators, who will jointly advise the graduate students. The budget includes support for travel and publicity allowing us to continue to build external support (See community involvement below). We have also budgeted for a smartphone testbed (Goal 3) consisting of a command-and-control server, 10 smartphone devices initially, and an additional set of 20 devices in years 2 and 3. Software will be disseminated through public software repositories such as GitHub, and will be used to host benchmarking results, a blog/wiki discussing best practices for evaluation, and the interface to the smartphone testbed. COMMUNITY INVOLVEMENT: As part of our planning grant (Award #1629791), PI Kennedy moderated a well-attended panel at ICDE 2017 entitled 'Small Data.' In addition to building interest in the community, two panel members: D. Richard Hipp and Eugene Wu are already using the preliminary results from the planning proposal.

1
Preproposal/letter_of_intent.txt Normal file
View File

@ -0,0 +1 @@
The world's 2 billion smartphones are a large part of the computing experience. A common requirement is persisting structured data, a task frequently performed by an embedded database like SQLite. Database performance can be a bottleneck, creating a poor user experience and causing unnecessary battery drain. Thus there are numerous opportunities for research on pocket-scale data management, or 'PocketData'. However PocketData research can be significantly more challenging than classical big-data research: (1) Smartphone hardware and operating systems self-regulate in unpredictable ways, making it difficult to obtain consistent, reproducible measurements. (2) Data accesses on a smartphone may be triggered in response to a variety of events, making it difficult to synthesize realistic workloads. (3) While small differences in hardware or operating system can lead to significant changes in system performance, it is not reasonable to expect researchers to obtain dozens of smartphones for testing purposes. This proposal aims to lower the barriers to entry for PocketData through three pieces of infrastructure: (1) We will develop a modular toolkit for PocketData performance measurement to enable consistent, reproducible results. (2) We will establish a benchmark for PocketData systems based on real-world traces gathered during the planning phase of this proposal. (3) We will deploy a testbed platform for smartphone researchers for consistent, reproducible, low-cost performance evaluation.

10
Preproposal/shortproposal.tex
View File

@ -26,11 +26,11 @@
}
\def\shortauthors{Ziarek, Kennedy, Nandi, Kul}
\def\submissiondate{2 Nov 2017}
\def\submissiondate{8 Jan 2019}
\def\theagency{CI-New; CISE Core Division: IIS}
\setlength{\lefttitle}{0.7\textwidth}
\setlength{\righttitle}{0.28\textwidth}
\def\thekeywords{databases, smartphones, reproducability}
\def\thekeywords{databases, smartphones, reproducibility}
\let\OLDthebibliography\thebibliography
\renewcommand\thebibliography[1]{
@ -47,11 +47,9 @@
{\small
\fromsite{Univ. at Buffalo, Dept. of Comp. Sci. \& Eng. \textbf{Lead Institution}}
{Oliver Kennedy, Co-PI: Lukasz Ziarek}\\
\fromsite{Ohio State University}
{Arnab Nandi}\\
\fromsite{Delaware State University}
{Gokhan Kul}\\
\textbf{Projected Budget Total}: \$750,000 across both sites \hfill \textit{(No External Collaborators)} \\
\textbf{Projected Budget Total}: \$955,000 across both sites \hfill \textit{(No External Collaborators)} \\
\textbf{Type}: \theagency; \textbf{Keywords}: \thekeywords; \hfill \submissiondate\\[-2mm]
}
\hrule
@ -74,7 +72,7 @@ Database performance can be a bottleneck~\cite{fitzpatrick-gdc10,yang-mobs13}, c
%that leads to apps beings sluggish or even failing with ``Application Not Responding" run-time errors
Mobile systems are rife with opportunities for data management research, as the precise chains of events that trigger poor performance can often be non-intuitive and hard to replicate.
For example, Android micro-manages CPU frequencies, reducing power use under light loads.
Counterintuitively, by being less efficient, a database can avoid triggering this feature and reduce latency.
Counter-intuitively, by being less efficient, a database can avoid triggering this feature and reduce latency.
Although there are numerous opportunities for research on pocket-scale data management, or \PocketData{}, it is significantly more challenging than classical ``big-data'' research:
\begin{enumerate*}