Improving resource management in virtualized data centers using application performance models

The rapid growth of virtualized data centers and cloud hosting services is making the management of physical resources such as CPU, memory, and I/O bandwidth in data center servers increasingly important. Server management now involves dealing with multiple dissimilar applications with varying Service-Level-Agreements (SLAs) and multiple resource dimensions. The multiplicity and diversity of resources and applications are rendering administrative tasks more complex and challenging. This thesis aimed to develop a framework and techniques that would help substantially reduce data center management complexity.^ We specifically addressed two crucial data center operations. First, we precisely estimated capacity requirements of client virtual machines (VMs) while renting server space in cloud environment. Second, we proposed a systematic process to efficiently allocate physical resources to hosted VMs in a data center. To realize these dual objectives, accurately capturing the effects of resource allocations on application performance is vital. The benefits of accurate application performance modeling are multifold. Cloud users can size their VMs appropriately and pay only for the resources that they need; service providers can also offer a new charging model based on the VMs performance instead of their configured sizes. As a result, clients will pay exactly for the performance they are actually experiencing; on the other hand, administrators will be able to maximize their total revenue by utilizing application performance models and SLAs. ^ This thesis made the following contributions. First, we identified resource control parameters crucial for distributing physical resources and characterizing contention for virtualized applications in a shared hosting environment. Second, we explored several modeling techniques and confirmed the suitability of two machine learning tools, Artificial Neural Network and Support Vector Machine, to accurately model the performance of virtualized applications. Moreover, we suggested and evaluated modeling optimizations necessary to improve prediction accuracy when using these modeling tools. Third, we presented an approach to optimal VM sizing by employing the performance models we created. Finally, we proposed a revenue-driven resource allocation algorithm which maximizes the SLA-generated revenue for a data center.^

Glycogen Synthase Kinase 3 Influences Cell Motility and Chemotaxis by Regulating Phosphatidylinositol 3 Kinase Localization in Dictyostelium discoideum

Glycogen Synthase Kinase 3 (GSK3), a serine/threonine kinase initially characterized in the context of glycogen metabolism, has been repeatedly realized as a multitasking protein that can regulate numerous cellular events in both metazoa and protozoa. I recently found GSK3 plays a role in regulating chemotaxis, a guided cell movement in response to an external chemical gradient, in one of the best studied model systems for chemotaxis - Dictyostelium discoideum. It was initially found that comparing to wild type cells, gsk3- cells showed aberrant chemotaxis with a significant decrease in both speed and chemotactic indices. In Dictyostelium, phosphatidylinositol 3,4,5-triphosphate (PIP3) signaling is one of the best characterized pathways that regulate chemotaxis. Molecular analysis uncovered that gsk3- cells suffer from high basal level of PIP3, the product of PI3K. Upon chemoattractant cAMP stimulation, wild type cells displayed a transient increase in the level of PIP3. In contrast, gsk3- cells exhibited neither significant increase nor adaptation. On the other hand, no aberrant dynamic of phosphatase and tensin homolog (PTEN), which antagonizes PI3K function, was observed. Upon membrane localization of PI3K, PI3K become activated by Ras, which will in turn further facilitate membrane localization of PI3K in an F-Actin dependent manner. The gsk3- cells treated with F-Actin inhibitor Latrunculin-A showed no significant difference in the PIP3 level. I also showed GSK3 affected the phosphorylation level of the localization domain of PI3K1 (PI3K1-LD). PI3K1-LD proteins from gsk3- cells displayed less phosphorylation on serine residues compared to that from wild type cells. When the potential GSK3 phosphorylation sites of PI3K1-LD were substituted with aspartic acids (Phosphomimetic substitution), its membrane localization was suppressed in gsk3- cells. When these serine residues of PI3K1-LD were substituted with alanine, aberrantly high level of membrane localization of the PI3K1-LD was monitored in wild type cells. Wild type, phosphomimetic, and alanine substitution of PI3K1-LD fused with GFP proteins also displayed identical localization behavior as suggested by the cell fraction studies. Lastly, I identified that all three potential GSK3 phosphorylation sites on PI3K1-LD could be phosphorylated in vitro by GSK3.

Improving Resource Management in Virtualized Data Centers using Application Performance Models

The rapid growth of virtualized data centers and cloud hosting services is making the management of physical resources such as CPU, memory, and I/O bandwidth in data center servers increasingly important. Server management now involves dealing with multiple dissimilar applications with varying Service-Level-Agreements (SLAs) and multiple resource dimensions. The multiplicity and diversity of resources and applications are rendering administrative tasks more complex and challenging. This thesis aimed to develop a framework and techniques that would help substantially reduce data center management complexity. We specifically addressed two crucial data center operations. First, we precisely estimated capacity requirements of client virtual machines (VMs) while renting server space in cloud environment. Second, we proposed a systematic process to efficiently allocate physical resources to hosted VMs in a data center. To realize these dual objectives, accurately capturing the effects of resource allocations on application performance is vital. The benefits of accurate application performance modeling are multifold. Cloud users can size their VMs appropriately and pay only for the resources that they need; service providers can also offer a new charging model based on the VMs performance instead of their configured sizes. As a result, clients will pay exactly for the performance they are actually experiencing; on the other hand, administrators will be able to maximize their total revenue by utilizing application performance models and SLAs. This thesis made the following contributions. First, we identified resource control parameters crucial for distributing physical resources and characterizing contention for virtualized applications in a shared hosting environment. Second, we explored several modeling techniques and confirmed the suitability of two machine learning tools, Artificial Neural Network and Support Vector Machine, to accurately model the performance of virtualized applications. Moreover, we suggested and evaluated modeling optimizations necessary to improve prediction accuracy when using these modeling tools. Third, we presented an approach to optimal VM sizing by employing the performance models we created. Finally, we proposed a revenue-driven resource allocation algorithm which maximizes the SLA-generated revenue for a data center.

The endocrine basis for reproductive life-history trade-offs during the previtellogenic resting stage in the yellow fever mosquito, Aedes Aegypti

Juvenile hormone (JH) is the central hormonal regulator of life-history trade-offs in many insects. In Aedes aegypti, JH regulates reproductive development after emergence. Little is known about JH's physiological functions after reproductive development is complete or JH's role in mediating life-history trade-offs. By examining the effect of hormones, nutrition, and mating on ovarian physiology during the previtellogenic resting stage, critical roles were determined for these factors in mediating life-history trade-offs and reproductive output. The extent of follicular resorption during the previtellogenic resting stage is dependent on nutritional quality. Feeding females a low quality diet during the resting stage causes the rate of follicular resorption to increase and reproductive output to decrease. Conversely, feeding females a high quality diet causes resorption to remain low. The extent of resorption can be increased by separating the ovaries from a source of JH or decreased by exogenous application of methoprene. Active caspases were localized to resorbing follicles indicating that an apoptosis-like mechanism participates in follicular resorption. Accumulations of neutral lipids and the accumulation of mRNA's integral to endocytosis and oocyte development such as the vitellogenin receptor (AaVgR), lipophorin receptor (AaLpRov), heavy-chain clathrin (AaCHC), and ribosomal protein L32 (rpL32) were also examined under various nutritional and hormonal conditions. The abundance of mRNA's and neutral lipid content increased within the previtellogenic ovary as mosquitoes were offered increasing sucrose concentrations or were treated with methoprene. These same nutritional and hormonal manipulations altered the extent of resorption after a blood meal indicating that the fate of follicles and overall fecundity depends, in part, on nutritional and hormonal status during the previtellogenic resting stage. Mating female mosquitoes also altered follicle quality and resorption similarly to nutrition or hormonal application and demonstrates that male accessory gland substances such as JH III passed to the female during copulation have a strong effect on ovarian physiology during the previtellogenic resting stage and can influence reproductive output. Taken together these results demonstrate that the previtellogenic resting stage is not an inactive period but is instead a period marked by extensive life-history and fitness trade-offs in response to nutrition, hormones and mating stimuli.

Manager-Customer Relationship in Food Service Commissary Operations

Documented reports of day-to-day decision-making in food service tend to emphasize technical aspects. However, this view does not represent completely the decision-making process managers go through. The author reports on the effect of the manager-customer relationship in decision-making by managers.

Are marginalized women being left behind? A population-based study of institutional deliveries in Karnataka, India

Background While India has made significant progress in reducing maternal mortality, attaining further declines will require increased skilled birth attendance and institutional delivery among marginalized and difficult to reach populations. Methods A population-based survey was carried out among 16 randomly selected rural villages in rural Mysore District in Karnataka, India between August and September 2008. All households in selected villages were enumerated and women with children 6 years of age or younger underwent an interviewer-administered questionnaire on antenatal care and institutional delivery. Results Institutional deliveries in rural areas of Mysore District increased from 51% to 70% between 2002 and 2008. While increasing numbers of women were accessing antenatal care and delivering in hospitals, large disparities were found in uptake of these services among different castes. Mothers belonging to general castes were almost twice as likely to have an institutional birth as compared to scheduled castes and tribes. Mothers belonging to other backward caste or general castes had 1.8 times higher odds (95% CI: 1.21, 2.89) of having an institutional delivery as compared to scheduled castes and tribes. In multivariable analysis, which adjusted for inter- and intra-village variance, Below Poverty Line status, caste, and receiving antenatal care were all associated with institutional delivery. Conclusion The results of the study suggest that while the Indian Government has made significant progress in increasing antenatal care and institutional deliveries among rural populations, further success in lowering maternal mortality will likely hinge on the success of NRHM programs focused on serving marginalized groups. Health interventions which target SC/ST may also have to address both perceived and actual stigma and discrimination, in addition to providing needed services. Strategies for overcoming these barriers may include sensitization of healthcare workers, targeted health education and outreach, and culturally appropriate community-level interventions. Addressing the needs of these communities will be critical to achieving Millennium Development Goal Five by 2015.

A novel signal processing method for intraoperative neurophysiological monitoring in spinal surgeries

Intraoperative neurophysiologic monitoring is an integral part of spinal surgeries and involves the recording of somatosensory evoked potentials (SSEP). However, clinical application of IONM still requires anywhere between 200 to 2000 trials to obtain an SSEP signal, which is excessive and introduces a significant delay during surgery to detect a possible neurological damage. The aim of this study is to develop a means to obtain the SSEP using a much less, twelve number of recordings. The preliminary step involved was to distinguish the SSEP with the ongoing brain activity. We first establish that the brain activity is indeed quasi-stationary whereas an SSEP is expected to be identical every time a trial is recorded. An algorithm was developed using Chebychev time windowing for preconditioning of SSEP trials to retain the morphological characteristics of somatosensory evoked potentials (SSEP). This preconditioning was followed by the application of a principal component analysis (PCA)-based algorithm utilizing quasi-stationarity of EEG on 12 preconditioned trials. A unique Walsh transform operation was then used to identify the position of the SSEP event. An alarm is raised when there is a 10% time in latency deviation and/or 50% peak-to-peak amplitude deviation, as per the clinical requirements. The algorithm shows consistency in the results in monitoring SSEP in up to 6-hour surgical procedures even under this significantly reduced number of trials. In this study, the analysis was performed on the data recorded in 29 patients undergoing surgery during which the posterior tibial nerve was stimulated and SSEP response was recorded from scalp. This method is shown empirically to be more clinically viable than present day approaches. In all 29 cases, the algorithm takes 4sec to extract an SSEP signal, as compared to conventional methods, which take several minutes. The monitoring process using the algorithm was successful and proved conclusive under the clinical constraints throughout the different surgical procedures with an accuracy of 91.5%. Higher accuracy and faster execution time, observed in the present study, in determining the SSEP signals provide a much improved and effective neurophysiological monitoring process.

Fabrication of a two-terminal super conducting device with a poled ferroelectric control layer

The discovery of High-Temperature Superconductors (HTSCs) has spurred the need for the fabrication of superconducting electronic devices able to match the performance of today's semiconductor devices. While there are several HTSCs in use today, YBaCuO7-x (YBCO) is the better characterized and more widely used material for small electronic applications. This thesis explores the fabrication of a Two-Terminal device with a superconductor and a painted on electrode as the terminals and a ferroelectric, BaTiO 3 (BTO), in between. The methods used to construct such a device and the challenges faced with the fabrication of a viable device will be examined. The ferroelectric layer of the devices that proved adequate for use were poled by the application of an electric field. Temperature Bias Poling used an applied field of 105V/cm at a temperature of approximately 135*C. High Potential Poling used an applied field of 106V/cm at room temperature (20*C). The devices were then tested for a change in their superconducting critical temperature, Tc. A shift of 1-2K in the Tc(onset) of YBCO was observed for Temperature Bias Poling and a shift of 2-6K for High Potential Poling. These are the first reported results of the field effect using BTO on YBCO. The mechanism involved in the shifting of Tc will be discussed along with possible applications.

Characterization of the poxAB Operon Encoding a Class D Carbapenemase in Pseudomonas aeruginosa,

Pseudomonas aeruginosa is a dreaded opportunistic pathogen that causes severe and often intractable infections in immunocompromised and critically ill patients. This bacterium is also the primary cause of fatal lung infections in patients with cystic fibrosis and a leading nosocomial pathogen responsible for nearly 10% of all hospital-acquired infections. P. aeruginosa is intrinsically recalcitrant to most classes of antibiotics and has the ability to acquire additional resistance during treatment. In particular, resistance to the widely used β-lactam antibiotics is frequently mediated by the expression of AmpC, a chromosomally encoded β-lactamase that is ubiquitously found in P. aeruginosa strains. This dissertation delved into the role of a recently reported chromosomal β-lactamase in P. aeruginosa called PoxB. To date, no detailed studies have addressed the regulation of poxB expression and its contribution to β-lactam resistance in P. aeruginosa. In an effort to better understand the role of this β-lactamase, poxB was deleted from the chromosome and expressed in trans from an IPTG-inducible promoter. The loss of poxB did not affect susceptibility. However, expression in trans in the absence of ampC rendered strains more resistant to the carbapenem β-lactams. The carbapenem-hydrolyzing phenotype was enhanced, reaching intermediate and resistant clinical breakpoints, in the absence of the carbapenem-specific outer membrane porin OprD. As observed for most class D β-lactamases, PoxB was only weakly inhibited by the currently available β-lactamase inhibitors. Moreover, poxB was shown to form an operon with the upstream located poxA, whose expression in trans decreased pox promoter (Ppox) activity suggesting autoregulation. The transcriptional regulator AmpR negatively controlled Ppox activity, however no direct interaction could be demonstrated. A mariner transposon library identified genes involved in the transport of polyamines as potential regulators of pox expression. Unexpectedly, polyamines themselves were able induce resistance to carbapenems. In summary, P. aeruginosa carries a chromosomal-encoded β-lactamase PoxB that can provide resistance against the clinically relevant carbapenems despite its narrow spectrum of hydrolysis and whose activity in vivo may be regulated by polyamines.

Design and analysis of micro-channel heat-exchanger embedded in Low Temperature Co-fire Ceramic (LTCC)

Increased device density, switching speeds of integrated circuits and decrease in package size is placing new demands for high power thermal-management. The convectional method of forced air cooling with passive heat sink can handle heat fluxes up-to 3-5W/cm2; however current microprocessors are operating at levels of 100W/cm2, This demands the usage of novel thermal-management systems. In this work, water-cooling systems with active heat sink are embedded in the substrate. The research involved fabricating LTCC substrates of various configurations - an open-duct substrate, the second with thermal vias and the third with thermal vias and free-standing metal columns and metal foil. Thermal testing was performed experimentally and these results are compared with CFD results. An overall thermal resistance for the base substrate is demonstrated to be 3.4oC/W-cm2. Addition of thermal vias reduces the effective resistance of the system by 7times and further addition of free standing columns reduced it by 20times.

Attitudes and behaviors of pre-adolescents with diabetes toward fat -modified foods

Fat modified foods are widely available and have the potential to help individuals with diabetes, including children, achieve a lower total fat and saturated fat intake. Sixty-three pre-adolescents (10-13 years) with insulin-dependent diabetes mellitus (IDDM or Type I), and 60 without diabetes (boys, n=54; girls, n=69) were tested to determine their beliefs and attitudes towards high-fat and reduced-fat foods. In addition, both children and parents were asked about the child's use of low fat foods i.e., how often the parent bought or encouraged their child to eat reduced-fat food; how strongly the doctor or dietitian promoted the use of reduced-fat foods, and the child's concern about dietary fat. In this study, preadolescents with diabetes were not more likely than those without diabetes to use fat-modified foods. Parental and health care practitioner encouragement is associated with greater use of these products by children.

Present-day Pattern Variations in the Central and Eastern Pacific El Nino

El Niño and the Southern Oscillation (ENSO) is a cycle that is initiated in the equatorial Pacific Ocean and is recognized on interannual timescales by oscillating patterns in tropical Pacific sea surface temperatures (SST) and atmospheric circulations. Using correlation and regression analysis of datasets that include SST’s and other interdependent variables including precipitation, surface winds, sea level pressure, this research seeks to quantify recent changes in ENSO behavior. Specifically, the amplitude, frequency of occurrence, and spatial characteristics (i.e. events with maximum amplitude in the Central Pacific versus the Eastern Pacific) are investigated. The research is based on the question; “Are the statistics of ENSO changing due to increasing greenhouse gas concentrations?” Our hypothesis is that the present-day changes in amplitude, frequency, and spatial characteristics of ENSO are determined by the natural variability of the ocean-atmosphere climate system, not the observed changes in the radiative forcing due to change in the concentrations of greenhouse gases. Statistical analysis, including correlation and regression analysis, is performed on observational ocean and atmospheric datasets available from the National Oceanographic and Atmospheric Administration (NOAA), National Center for Atmospheric Research (NCAR) and coupled model simulations from the Coupled Model Inter-comparison Project (phase 5, CMIP5). Datasets are analyzed with a particular focus on ENSO over the last thirty years. Understanding the observed changes in the ENSO phenomenon over recent decades has a worldwide significance. ENSO is the largest climate signal on timescales of 2 - 7 years and affects billions of people via atmospheric teleconnections that originate in the tropical Pacific. These teleconnections explain why changes in ENSO can lead to climate variations in areas including North and South America, Asia, and Australia. For the United States, El Niño events are linked to decreased number of hurricanes in the Atlantic basin, reduction in precipitation in the Pacific Northwest, and increased precipitation throughout the southern United Stated during winter months. Understanding variability in the amplitude, frequency, and spatial characteristics of ENSO is crucial for decision makers who must adapt where regional ecology and agriculture are affected by ENSO.

Enhanced Alphabet Knowledge in Preschool Children

Providing a child the opportunity to succeed in school is a main worry of parents and teachers. When children are able to connect letters with their corresponding sounds allows for literacy to grow. Using Enhanced Alphabet Knowledge (EAK) instruction will allow children to evolve their literacy skills by connecting.

Mixed Spatial and Nonspatial Problems in Location Based Services

With hundreds of millions of users reporting locations and embracing mobile technologies, Location Based Services (LBSs) are raising new challenges. In this dissertation, we address three emerging problems in location services, where geolocation data plays a central role. First, to handle the unprecedented growth of generated geolocation data, existing location services rely on geospatial database systems. However, their inability to leverage combined geographical and textual information in analytical queries (e.g. spatial similarity joins) remains an open problem. To address this, we introduce SpsJoin, a framework for computing spatial set-similarity joins. SpsJoin handles combined similarity queries that involve textual and spatial constraints simultaneously. LBSs use this system to tackle different types of problems, such as deduplication, geolocation enhancement and record linkage. We define the spatial set-similarity join problem in a general case and propose an algorithm for its efficient computation. Our solution utilizes parallel computing with MapReduce to handle scalability issues in large geospatial databases. Second, applications that use geolocation data are seldom concerned with ensuring the privacy of participating users. To motivate participation and address privacy concerns, we propose iSafe, a privacy preserving algorithm for computing safety snapshots of co-located mobile devices as well as geosocial network users. iSafe combines geolocation data extracted from crime datasets and geosocial networks such as Yelp. In order to enhance iSafe's ability to compute safety recommendations, even when crime information is incomplete or sparse, we need to identify relationships between Yelp venues and crime indices at their locations. To achieve this, we use SpsJoin on two datasets (Yelp venues and geolocated businesses) to find venues that have not been reviewed and to further compute the crime indices of their locations. Our results show a statistically significant dependence between location crime indices and Yelp features. Third, review centered LBSs (e.g., Yelp) are increasingly becoming targets of malicious campaigns that aim to bias the public image of represented businesses. Although Yelp actively attempts to detect and filter fraudulent reviews, our experiments showed that Yelp is still vulnerable. Fraudulent LBS information also impacts the ability of iSafe to provide correct safety values. We take steps toward addressing this problem by proposing SpiDeR, an algorithm that takes advantage of the richness of information available in Yelp to detect abnormal review patterns. We propose a fake venue detection solution that applies SpsJoin on Yelp and U.S. housing datasets. We validate the proposed solutions using ground truth data extracted by our experiments and reviews filtered by Yelp.