Wenyao Zhang

Cost-Aware Cooperative Resource Provisioning for Heterogeneous Workloads in Data Centers

Recent cost analysis shows that the server cost still dominates the total cost of high-scale data centers or cloud systems. In this paper, we argue for a new twist on the classical resource provisioning problem: heterogeneous workloads are a fact of life in large-scale data centers, and current resource provisioning solutions do not act upon this heterogeneity. Our contributions are threefold: first, we propose a cooperative resource provisioning solution, and take advantage of differences of heterogeneous workloads so as to decrease their peak resources consumption under competitive conditions; second, for four typical heterogeneous workloads: parallel batch jobs, web servers, search engines, and MapReduce jobs, we build an agile system PhoenixCloud that enables cooperative resource provisioning; and third, we perform a comprehensive evaluation for both real and synthetic workload traces. Our experiments show that our solution could save the server cost aggressively with respect to the noncooperative solutions that are widely used in state-of-the-practice hosting data centers or cloud systems: for example, EC2, which leverages the statistical multiplexing technique, or RightScale, which roughly implements the elastic resource provisioning technique proposed in related state-of-the-art work.

Topology-driven streamline seeding for 2D vector field visualization

This paper presents a novel method of streamline placement for 2D vector field. In this method, the topological skeleton of underlying field is firstly extracted and used as initial streamlines. Initial streamlines segment the field into topological areas. Additional streamlines are then seeded at the center of topological areas in a recursive way, until there is no any valid empty area. To implement this method efficiently, a boundary extending of vector field is used to simplify the extraction of topology. And a virtual control grid superposed on the field is used to model topological areas approximately, and control the density of streamlines as well as their length. Our method focuses on the topology of vector field while keeping streamlines evenly-spaced as possible. But it can still run without the initial topological skeleton. Test results show that our method can achieve high quality of streamline placement.

Multiresolution streamline placement based on control grids

Multiresolution streamline placement is useful for interactive visualization of flow fields. In this paper, a new method is proposed to build placements of streamlines in different resolutions for 2D and 3D flow fields. This method is developed from a basic streamline placement algorithm that takes control grids to place streamlines in 2D flow fields. The process of multiresolution analysis starts from a sparse placement, which is obtained with a coarse grid according to the basic placement algorithm. The sparse placement is then iteratively refined by elongating all existing streamlines and adding new streamlines under the control of a finer grid. By this means, a series of placements can be built in different resolutions, each of which reuses its previously placed streamlines at increasing density levels. This multiresolution placement for 2D flow fields is further extended to 3D flow fields by changing the basic placement algorithm to fit 3D space. Test results show that the basic streamline placement algorithm is very fast, so the proposed multiresolution analysis can be done for both 2D and 3D flow fields efficiently.

A Streamline Placement Method Highlighting Flow Field Topology

Streamline is an important way to visualize flow fields. In this paper, we propose a new streamline placement method. This method seeds streamlines at the centers of sub-fields that are enclosed by existing streamlines or field boundaries. When a streamline is seeded in a sub-field, the sub-field is generally partitioned into one or more small ones. We then continue the placement in the new sub-fields until there are no any empty ones left. The original field is treated as the first unique sub-field. For the efficiency of implementation, a sub-field is modeled as a four-connected empty area in a predefined control grid. When the control grid is properly set, test results show that, our method can generate almost evenly-spaced streamlines everywhere except at the neighborhood of topological components such as critical points and separatrices. Therefore, the topological structure of flow fields is highlighted by the varying density of streamlines without any additional overhead on analyzing the topology. This is the main advantage that is not owned by other placement methods. In addition, the cost of computation in our method is rather low.

A feature-emphasized clustering method for 2D vector field

Large-scale vector data produce the vector field clustering in flow visualization. To emphasize essential flow features, a new clustering method for 2D vector fields is proposed in this paper. With this method, the vector field is firstly initialized as a cluster, which is then iteratively divided into a hierarchy of clusters. During the iteration, clusters are segmented with streamlines instead of straight lines. This change enables it to emphasize flow features, since streamlines are consistent with flow behaviors, and clusters shaped by streamlines are aligned to the underlying flow. It is easy to capture flow patterns and features from resulting clusters. Moreover, our method improves representative vectors of clusters, leading to a more efficient approximation to the original field. Test results show that it is superior to other similar methods in terms of preserving flow features and approximating vector fields.

Parallel Streamline Placement for 2D Flow Fields

Parallel streamline placement is still an open problem in flow visualization. In this paper, we propose an innovative method to place streamlines in parallel for 2D flow fields. This method is based on our proposed concept of local tracing areas (LTAs). An LTA is defined as a subdomain enclosed by streamlines and/or field borders, where the tracing of streamlines are localized. Given a flow field, it is initialized as an LTA, which is later recursively partitioned into hierarchical LTAs. Streamlines are placed within different LTAs simultaneously and independently. At the same time, to control the density of streamlines, each streamline is associated with an isolation zone and a saturation zone, both of which are center aligned with the streamline but have different widths. None of streamlines can trace into isolation zones of others. And new streamlines are only seeded within valid seeding areas (VSAs) that are enclosed by saturation zones and/or field borders. To implement the parallel strategy and the density control, a cell-based modeling is devised to describe isolation zones and LTAs as well as saturation zones and VSAs. With the help of these cell-based models, a heuristic seeding strategy is proposed to seed streamlines within irregular LTAs, and a cell-marking technique is used to control the seeding and tracing of streamlines. Test results show that the placement method can achieve highly parallel performance on shared memory systems without losing the quality of placements.

Streamline-based Visualization of 3D Explosion Fields

The visualization of explosion fields is an important issue in explosion science and technology. This paper presents a streamline-based method to visualize 3D explosion fields. Given the velocity data in a 3D explosion field, is surfaces of the magnitude of velocity are firstly extracted. Seeds of streamlines are then uniformly distributed on the is surfaces. And streamlines are generated with the placed seeds. By this way, the density of streamlines is controlled by a predefined value. When the is surfaces and the density value are properly set, the coverage of streamlines can be sufficient. To reduce visual cluttering, a technique of line illumination is used to render streamlines. Moreover, a color coding technique is applied to visualize the magnitude of velocity and distinguish the distribution of media in 3D explosion fields. Test results show that the method proposed in this paper can achieve effective visualization of 3D explosion fields. The visual clarity and the spatial perception are improved in the final visualized images.

Multiresolution Streamline Placement for 2D Flow Fields

Streamline is a popular choice to visualize flow fields, but it not easy to determine the number and the distribution of streamlines. In this paper, a multiresolution streamline placement method is presented to visualize 2D flow fields at different levels of detail. Each level has different number of streamlines that are uniformly distributed in the field at different resolutions. This method is implemented by constructing a hierarchy of control grids to manipulate the procedure of placement. The top level of the hierarchy is the coarsest control grid whose cell size is predefined. Using this grid and a centroid-based seeding strategy, a sparse placement is obtained, where streamlines are evenly-spaced. Then, this sparse placement is changed into the next lower level by adding some new streamlines within the gaps of all existing streamlines under the control of a finer control grid. Although the total number of streamlines is increased, the original uniformity is not destroyed. The resolution and the density of streamlines are both improved from one level to its next lower level. This refinement between adjacent levels can be repeated until a suitable placement is obtained. Corresponding to the control grids, a hierarchy of streamline placements with different resolutions is also built up during the multiresolution analysis. It gives an opportunity to select suitable streamline placements with desired resolutions. And it also provides a new way to have an insight into the flow field at different scales. Test results show that this multiresolution streamline placement is valid and efficient.

Joint Visualization of UKF Tractography Data

Tractography methods provide ways to explore white mater in brain. UKF Tractography is a promising one in processing cross fibers and edema regions. In order to get more insight into UKF tractography, we present a joint visualization scheme for UKF tractography data, which integrates and visualizes fiber tracts, diffusion tensors, and multiple tensor measures at the same time. In the scheme, a new kind of tensor glyphs called cylingons, which is similar to cylinder but has more lateral faces, is designed to hold multiple scalar measures through the color mapping of faces. Cylingons are then combined with fiber tracts to provide global and local views under the control of different visualizing options. Initial tests and applications show that our approach can do most work of visual analysis for UKF tractography data, and help to get something that is worthy of further study.

An Improved Fiber Tracking Method for Crossing Fibers

Fiber tracking is a basic task in analyzing data obtained by diffusion tensor magnetic resonance imaging (DT-MRI). In order to get a better tracking result for crossing fibers with noise, an improved fiber tracking method is proposed in this paper. The method is based on the framework of Bayesian fiber tracking, but improves its ability to deal with crossing fibers, by introducing the high order tensor (HOT) model as well as a new fiber direction selection strategy. In this method, orientation distribution function is first obtained from HOT model, and then used as the likelihood probability to control fiber tracing. On this basis, the direction in candidates that has the smallest change relative to current two previous directions is selected as the next tracing direction. By this means, our method achieves better performance in processing crossing fibers.