Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
2Activity
0 of .
Results for:
No results containing your search query
P. 1
NWU-EECS-07-04

NWU-EECS-07-04

Ratings: (0)|Views: 32 |Likes:

More info:

Categories:Types, Research
Published by: eecs.northwestern.edu on Jun 02, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

06/02/2010

pdf

text

original

 
 
Electrical Engineering and Computer Science Department
Technical ReportNWU-EECS-07-04July 26, 2007Human-driven Optimization
Bin Lin
Abstract
The optimization problems associated with adaptive and autonomic computing systemsare often difficult to pose well and solve efficiently. A key challenge is that for manyapplications, particularly interactive applications, the user or developer is unlikely or unable to provide either the objective function
 f  
, or constraints. It is a key problemencountered broadly in adaptive and autonomic computing.This dissertation argues for using
human-driven optimization
techniques to solveoptimization problems. In particular, it consists of two core ideas. In
human-drivenspecification
, we use direct human input from users to pose specific optimization problems, namely to determine the objective function
 f  
and expose hidden constraints.Once we have a well-specified problem, we are left with the need to search for a solutionin a very large solution space. In
human-driven search
, we use direct human input toguide the search for a good solution, a valid configuration x that optimizes
 f(x)
.My research happens in three contexts. The main context is the Virtuoso system for utility and grid computing based on virtual machines (VMs) interconnected with overlaynetworks. Virtuoso provides instances of optimization problems and a framework for evaluating solutions to them. In particular, the virtual execution environment of Virtuosomakes possible low-level, application- and developer-independent adaptationmechanisms such as CPU reservations, VM migration, overlay topology configurationand routing, and network reservations. The high-level optimization problem in Virtuosois how to dynamically optimize, at run-time, the performance of existing, unmodifieddistributed applications running on existing, unmodified operating systems. Theseapplications can be batch, batch parallel and interactive applications. The second contextof my research is power management for laptops. Existing Dynamic voltage andFrequency Scaling (DVFS) techniques are commonly used to reduce power consumption
 
 but they are conservative and pessimistic about both the user and the processor. How tofurther prolong battery life and reduce heat dissipation is the problem that is addressed.The last context of my research is IT configuration, a process whereby individualcomponents are assembled and adjusted to construct a working solution. Visiblecomplexity—of setting configuration knobs, installing and updating software, diagnosingand repairing problems, and so on—is a challenge for IT. How to reduce such complexityis a challenging problem. As the first step, I explored the decision complexity presentedto the non-expert system administrator, which represents a significant part of the wholeIT complexity.To show the feasibility and effectiveness of my techniques, in this dissertation, I describehow I address increasingly difficult optimization problems in the Virtuoso context usinghuman-driven specification or search. Those problems cover single machine CPUscheduling, multiple machine CPU scheduling, and multiple machine VM mapping for interactive (desktop), batch and batch parallel applications. I also present how I applyhuman-driven techniques to solving power management problems on laptop computers.In general, solving each of these problems involves the design and development of systems mechanisms, adaptive algorithms and user interfaces. I evaluate each element of my work through a user study. I also discuss my work on modeling user decisioncomplexity in IT configuration systems, as the first step towards applying human-driventechniques in that domain.
This work is in part supported by the NSF (awards ANI-0093221, ANI-0301108, and EIA-0224449), theDOE via ORNL, and by gifts from VMware, Dell, and Symantec.
Keywords: Adaptive systems; Human-computer interaction; Resource virtualization; Power management
 
NORTHWESTERN UNIVERSITYHuman-driven OptimizationA DISSERTATIONSUBMITTED TO THE GRADUATE SCHOOLIN PARTIAL FULFILLMENT OF THE REQUIREMENTSfor the degreeDOCTOR OF PHILOSOPHYField of Computer ScienceByBin LinEVANSTON, ILLINOISDecember 2007

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->