Managing server energy and reducing operational cost for online service providers

The past decade has seen the energy consumption in servers and Internet Data Centers (IDCs) skyrocket. A recent survey estimated that the worldwide spending on servers and cooling have risen to above $30 billion and is likely to exceed spending on the new server hardware . The rapid rise in energy consumption has posted a serious threat to both energy resources and the environment, which makes green computing not only worthwhile but also necessary. This dissertation intends to tackle the challenges of both reducing the energy consumption of server systems and by reducing the cost for Online Service Providers (OSPs). Two distinct subsystems account for most of IDC’s power: the server system, which accounts for 56% of the total power consumption of an IDC, and the cooling and humidifcation systems, which accounts for about 30% of the total power consumption. The server system dominates the energy consumption of an IDC, and its power draw can vary drastically with data center utilization. In this dissertation, we propose three models to achieve energy effciency in web server clusters: an energy proportional model, an optimal server allocation and frequency adjustment strategy, and a constrained Markov model. The proposed models have combined Dynamic Voltage/Frequency Scaling (DV/FS) and Vary-On, Vary-off (VOVF) mechanisms that work together for more energy savings. Meanwhile, corresponding strategies are proposed to deal with the transition overheads. We further extend server energy management to the IDC’s costs management, helping the OSPs to conserve, manage their own electricity cost, and lower the carbon emissions. We have developed an optimal energy-aware load dispatching strategy that periodically maps more requests to the locations with lower electricity prices. A carbon emission limit is placed, and the volatility of the carbon offset market is also considered. Two energy effcient strategies are applied to the server system and the cooling system respectively. With the rapid development of cloud services, we also carry out research to reduce the server energy in cloud computing environments. In this work, we propose a new live virtual machine (VM) placement scheme that can effectively map VMs to Physical Machines (PMs) with substantial energy savings in a heterogeneous server cluster. A VM/PM mapping probability matrix is constructed, in which each VM request is assigned with a probability running on PMs. The VM/PM mapping probability matrix takes into account resource limitations, VM operation overheads, server reliability as well as energy effciency. The evolution of Internet Data Centers and the increasing demands of web services raise great challenges to improve the energy effciency of IDCs. We also express several potential areas for future research in each chapter.

High-resolution morphological and biochemical imaging of articular cartilage of the ankle joint at 3.0 T using a new dedicated phased array coil: in vivo reproducibility study

OBJECTIVE: The objective of this study was to evaluate the feasibility and reproducibility of high-resolution magnetic resonance imaging (MRI) and quantitative T2 mapping of the talocrural cartilage within a clinically applicable scan time using a new dedicated ankle coil and high-field MRI. MATERIALS AND METHODS: Ten healthy volunteers (mean age 32.4 years) underwent MRI of the ankle. As morphological sequences, proton density fat-suppressed turbo spin echo (PD-FS-TSE), as a reference, was compared with 3D true fast imaging with steady-state precession (TrueFISP). Furthermore, biochemical quantitative T2 imaging was prepared using a multi-echo spin-echo T2 approach. Data analysis was performed three times each by three different observers on sagittal slices, planned on the isotropic 3D-TrueFISP; as a morphological parameter, cartilage thickness was assessed and for T2 relaxation times, region-of-interest (ROI) evaluation was done. Reproducibility was determined as a coefficient of variation (CV) for each volunteer; averaged as root mean square (RMSA) given as a percentage; statistical evaluation was done using analysis of variance. RESULTS: Cartilage thickness of the talocrural joint showed significantly higher values for the 3D-TrueFISP (ranging from 1.07 to 1.14 mm) compared with the PD-FS-TSE (ranging from 0.74 to 0.99 mm); however, both morphological sequences showed comparable good results with RMSA of 7.1 to 8.5%. Regarding quantitative T2 mapping, measurements showed T2 relaxation times of about 54 ms with an excellent reproducibility (RMSA) ranging from 3.2 to 4.7%. CONCLUSION: In our study the assessment of cartilage thickness and T2 relaxation times could be performed with high reproducibility in a clinically realizable scan time, demonstrating new possibilities for further investigations into patient groups.

Predicting the Mechanical Properties of Carbon-Based Materials Using Molecular Dynamics

The development of innovative carbon-based materials can be greatly facilitated by molecular modeling techniques. Although molecular modeling has been used extensively to predict elastic properties of materials, modeling of more complex phenomenon such as fracture has only recently been possible with the development of new force fields such as ReaxFF, which is used in this work. It is not fully understood what molecular modeling parameters such as thermostat type, thermostat coupling, time step, system size, and strain rate are required for accurate modeling of fracture. Selection of modeling parameters to model fracture can be difficult and non-intuitive compared to modeling elastic properties using traditional force fields, and the errors generated by incorrect parameters may be non-obvious. These molecular modeling parameters are systematically investigated and their effects on the fracture of well-known carbon materials are analyzed. It is determined that for coupling coefficients of 250 fs and greater do not result in substantial differences in the stress-strain response of the materials using any thermostat type. A time step of 0.5 fs of smaller is required for accurate results. Strain rates greater than 2.2 ns-1 are sufficient to obtain repeatable results with slower strain rates for the materials studied. The results of this study indicate that further refinement of the Chenoweth parameter set is required to accurately predict the mechanical response of carbon-based systems. The ReaxFF has been used extensively to model systems in which bond breaking and formation occur. In particular ReaxFF has been used to model reactions of small molecules. Some elastic and fracture properties have been successfully modeled using ReaxFF in materials such as silicon and some metals. However, it is not clear if current parameterizations for ReaxFF are able to accurately reproduce the elastic and fracture properties of carbon materials. The stress-strain response of a new ReaxFF parameterization is compared to the previous parameterization and density functional theory results for well-known carbon materials. The new ReaxFF parameterization makes xv substantial improvements to the predicted mechanical response of carbon materials, and is found to be suitable for modeling the mechanical response of carbon materials. Finally, a new material composed of carbon nanotubes within an amorphous carbon (AC) matrix is modeled using the ReaxFF. Various parameters that may be experimentally controlled are investigated such as nanotube bundling, comparing multi-walled nanotube with single-walled nanotubes, and degree of functionalization of the nanotubes. Elastic and fracture properties are investigated for the composite systems and compared to results of pure-nanotube and pure-AC models. It is found that the arrangement of the nanotubes and degree of crosslinking may substantially affect the properties of the systems, particularly in the transverse directions.

Development and characterization of fluorescent pH sensors based on porous silica and hydrogel support matrices

The hydrogen ion activity (pH) is a very important parameter in environment monitoring, biomedical research and other applications. Optical pH sensors have several advantages over traditional potentiometric pH measurement, such as high sensitivity, no need of constant calibration, easy for miniaturization and possibility for remote sensing. Several pH indicators has been successfully immobilized in three different solid porous materials to use as pH sensing probes. The fluorescent pH indicator fluorescein-5-isothiocyanate (FITC) was covalently bound onto the internal surface of porous silica (pore size ~10 nm) and retained its pH sensitivity. The excited state pK* a of FITC in porous silica (5.58) was slightly smaller than in solution (5.68) due to the free silanol groups (Si-OH) on the silica surface. The pH sensitive range for this probe is pH 4.5 - 7.0 with an error less than 0.1 pH units. The probe response was reproducible and stable for at least four month, stored in DI water, but exhibit a long equilibrium of up to 100 minutes. Sol-gel based pH sensors were developed with immobilization of two fluorescent pH indicators fluorescein-5-(and-6)-sulfonic acid, trisodium salt (FS) and 8-hydroxypyrene- 1,3,6-trisulfonic acid (HPTS) through physical entrapment. Prior to immobilization, the indicators were ion-paired with a common surfactant hexadecyltrimethylammonium bromide (CTAB) in order to prevent leaching. The sol-gel films were synthesized through the hydrolysis of two different precursors, ethyltriethoxysilane (ETEOS) and 3- glycidoxypropyltrimethoxysilane (GPTMS) and deposited on a quartz slide through spin coating. The pK a of the indicators immobilized in sol-gel films was much smaller than in solutions due to silanol groups on the inner surface of the sol-gel films and ammonium groups from the surrounding surfactants. Unlike in solution, the apparent pK a of the indicators in sol-gel films increased with increasing ionic strength. The equilibrium time for these sensors was within 5 minutes (with film thickness of ~470 nm). Polyethylene glycol (PEG) hydrogel was of interest for optical pH sensor development because it is highly proton permeable, transparent and easy to synthesize. pH indicators can be immobilized in hydrogel through physical entrapment and copolymerization. FS and HPTS ion-pairs were physically entrapped in hydrogel matrix synthesized via free radical initiation. For covalent immobilization, three indicators, 6,8-dihydroxypyrene-1,3- disulfonic acid (DHPDS), 2,7-dihydroxynaphthalene-3,6-disulfonic acid (DHNDS) and cresol red were first reacted with methacrylic anhydride (MA) to form methacryloylanalogs for copolymerization. These hydrogels were synthesized in aqueous solution with a redox initiation system. The thickness of the hydrogel film is controlled as ~ 0.5 cm and the porosity can be adjusted with the percentage of polyethylene glycol in the precursor solutions. The pK a of the indicators immobilized in the hydrogel both physically and covalently were higher than in solution due to the medium effect. The sensors are stable and reproducible with a short equilibrium time (less than 4 minutes). In addition, the color change of cresol red immobilized hydrogel is vivid from yellow (acidic condition) to purple (basic condition). Due to covalently binding, cresol red was not leaching out from the hydrogel, making it a good candidate of reusable "pH paper".

An Evaluation of Potential Occupational Exposure to Asbestiform Amphiboles near a Former Vermiculite Mine

Amphibole asbestos (AA) has been detected on the surface of tree bark in forests neighboring an abandoned vermiculite mine near Libby, Montana. In the present study, simulations were performed to assess potential AA exposure associated with United States Department of Agriculture Forest Service (FS) occupational activities. Bark samples were collected prior, and personal breathing zone (PBZ) and Tyvek clothing wipe samples were collected during and immediately after trials that simulated FS activities. Transmission electron microscopy (TEM) analyses revealed AA bark concentrations up to 15 million structures per square centimeter (s/cm2). AA was detected in 25% of the PBZ TEM samples. AA was detected on wipe samples collected from all activities evaluated. This research demonstrates the potential for airborne exposure and transport of AA in the Kootenai National Forest. These findings are especially relevant to those that work in the area and to the general public who may conduct recreational activities.

Congenital thrombotic thrombocytopenic purpura caused by new compound heterozygous mutations of the ADAMTS13 gene

Upshaw-Schulman syndrome (USS) is due to severe congenital deficiency of von Willebrand factor (VWF)-cleaving protease ADAMTS13 (a disintegrin and metalloprotease with thrombospondin type 1 domains, nr 13) activity resulting in the presence of unusually large forms of VWF in the circulation, causing intravascular platelet clumping and thrombotic microangiopathy. Our patient, a 26-year-old man, had attacks of thrombotic thrombocytopenic purpura (TTP) with thrombocytopenia and a urine dipstick positive for hemoglobin (4+), often as the only sign of hemolytic activity. He had ADAMTS13 activity of <1% of normal plasma without the presence of inhibitors of ADAMTS13. ADAMTS13 deficiency was caused by two new mutations of the ADAMTS13 gene: a deletion of a single nucleotide in exon17 (c. 2042 delA) leading to a frameshift (K681C fs X16), and a missense mutation in exon 25 (c.3368G>A) leading to p.R1123H. This case report confirms the importance of the analysis of the ADAMTS13 activity and its inhibitor in patients who have episodes of TTP, with a very low platelet count and sometimes without the classic biochemical signs of hemolysis.

Ultrafast Relaxation Dynamics of Osmium-Polypyridine Complexes in Solution

We present steady-state absorption and emission spectroscopy and femtosecond broadband photoluminescence up-conversion spectroscopy studies of the electronic relaxation of Os(dmbp)3 (Os1) and Os(bpy)2(dpp) (Os2) in ethanol, where dmbp is 4,4′-dimethyl-2,2′-biypridine, bpy is 2,2′-biypridine, and dpp is 2,3-dipyridyl pyrazine. In both cases, the steady-state phosphorescence is due to the lowest 3MLCT state, whose quantum yield we estimate to be ≤5.0 × 10–3. For Os1, the steady-state phosphorescence lifetime is 25 ns. In both complexes, the photoluminescence excitation spectra map the absorption spectrum, pointing to an excitation wavelength-independent quantum yield. The ultrafast studies revealed a short-lived (≤100 fs) fluorescence, which stems from the lowest singlet metal-to-ligand-charge-transfer (1MLCT) state and decays by intersystem crossing to the manifold of 3MLCT states. In addition, Os1 exhibits a 50 ps lived emission from an intermediate triplet state at an energy 2000 cm–1 above that of the long-lived (25 ns) phosphorescence. In Os2, the 1MLCT–3MLCT intersystem crossing is faster than that in Os1, and no emission from triplet states is observed other than the lowest one. These observations are attributed to a higher density of states or a smaller energy spacing between them compared with Os1. They highlight the importance of the energetics on the rate of intersystem crossing.

Real-time observation of the charge-transfer-to-solvent dynamics

Intermolecular electron-transfer reactions have a crucial role in biology, solution chemistry and electrochemistry. The first step of such reactions is the expulsion of the electron to the solvent, whose mechanism is determined by the structure and dynamical response of the latter. Here we visualize the electron transfer to water using ultrafast fluorescence spectroscopy with polychromatic detection from the ultraviolet to the visible region, upon photo-excitation of the so-called charge transfer to solvent states of aqueous iodide. The initial emission is short lived (~60 fs) and it relaxes to a broad distribution of lower-energy charge transfer to solvent states upon rearrangement of the solvent cage. This distribution reflects the inhomogeneous character of the solvent cage around iodide. Electron ejection occurs from the relaxed charge transfer to solvent states with lifetimes of 100–400 fs that increase with decreasing emission energy.