Host and network optimizations for performance enhancement and energy efficiency in data center networks

Modern data centers host hundreds of thousands of servers to achieve economies of scale. Such a huge number of servers create challenges for the data center network (DCN) to provide proportionally large bandwidth. In addition, the deployment of virtual machines (VMs) in data centers raises the requirements for efficient resource allocation and find-grained resource sharing. Further, the large number of servers and switches in the data center consume significant amounts of energy. Even though servers become more energy efficient with various energy saving techniques, DCN still accounts for 20% to 50% of the energy consumed by the entire data center. The objective of this dissertation is to enhance DCN performance as well as its energy efficiency by conducting optimizations on both host and network sides. First, as the DCN demands huge bisection bandwidth to interconnect all the servers, we propose a parallel packet switch (PPS) architecture that directly processes variable length packets without segmentation-and-reassembly (SAR). The proposed PPS achieves large bandwidth by combining switching capacities of multiple fabrics, and it further improves the switch throughput by avoiding padding bits in SAR. Second, since certain resource demands of the VM are bursty and demonstrate stochastic nature, to satisfy both deterministic and stochastic demands in VM placement, we propose the Max-Min Multidimensional Stochastic Bin Packing (M3SBP) algorithm. M3SBP calculates an equivalent deterministic value for the stochastic demands, and maximizes the minimum resource utilization ratio of each server. Third, to provide necessary traffic isolation for VMs that share the same physical network adapter, we propose the Flow-level Bandwidth Provisioning (FBP) algorithm. By reducing the flow scheduling problem to multiple stages of packet queuing problems, FBP guarantees the provisioned bandwidth and delay performance for each flow. Finally, while DCNs are typically provisioned with full bisection bandwidth, DCN traffic demonstrates fluctuating patterns, we propose a joint host-network optimization scheme to enhance the energy efficiency of DCNs during off-peak traffic hours. The proposed scheme utilizes a unified representation method that converts the VM placement problem to a routing problem and employs depth-first and best-fit search to find efficient paths for flows.

Assessment of natural resource conditions in and adjacent to Biscayne National Park. Natural Resource Report NPS/BISC/NRR—2012/598

This report is an assessment of the conditions of natural resources in Biscayne National Park (BNP) based on the compilation, review and evaluation of existing information on the Park’s natural resources. This review evaluates threats and stressors, and is intended to improve understanding of BNP resources to help guide Park management to address the identified threats, which are supported by enhanced data collection, research and assessment efforts. The report is focused on broad resource components, namely terrestrial resources and aquatic systems including: wetlands, canals, bay waters, marine/reef areas and ground waters. Biotic and abiotic resource components are considered in the review.

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.^

Design and implementation of a web based shop floor control system to maintain schedule feasibility in a dynamic job shop environment

The effective control of production activities in dynamic job shop with predetermined resource allocation for all the jobs entering the system is a unique manufacturing environment, which exists in the manufacturing industry. In this thesis a framework for an Internet based real time shop floor control system for such a dynamic job shop environment is introduced. The system aims to maintain the schedule feasibility of all the jobs entering the manufacturing system under any circumstance. The system is capable of deciding how often the manufacturing activities should be monitored to check for control decisions that need to be taken on the shop floor. The system will provide the decision maker real time notification to enable him to generate feasible alternate solutions in case a disturbance occurs on the shop floor. The control system is also capable of providing the customer with real time access to the status of the jobs on the shop floor. The communication between the controller, the user and the customer is through web based user friendly GUI. The proposed control system architecture and the interface for the communication system have been designed, developed and implemented.

Host and Network Optimizations for Performance Enhancement and Energy Efficiency in Data Center Networks

Modern data centers host hundreds of thousands of servers to achieve economies of scale. Such a huge number of servers create challenges for the data center network (DCN) to provide proportionally large bandwidth. In addition, the deployment of virtual machines (VMs) in data centers raises the requirements for efficient resource allocation and find-grained resource sharing. Further, the large number of servers and switches in the data center consume significant amounts of energy. Even though servers become more energy efficient with various energy saving techniques, DCN still accounts for 20% to 50% of the energy consumed by the entire data center. The objective of this dissertation is to enhance DCN performance as well as its energy efficiency by conducting optimizations on both host and network sides. First, as the DCN demands huge bisection bandwidth to interconnect all the servers, we propose a parallel packet switch (PPS) architecture that directly processes variable length packets without segmentation-and-reassembly (SAR). The proposed PPS achieves large bandwidth by combining switching capacities of multiple fabrics, and it further improves the switch throughput by avoiding padding bits in SAR. Second, since certain resource demands of the VM are bursty and demonstrate stochastic nature, to satisfy both deterministic and stochastic demands in VM placement, we propose the Max-Min Multidimensional Stochastic Bin Packing (M3SBP) algorithm. M3SBP calculates an equivalent deterministic value for the stochastic demands, and maximizes the minimum resource utilization ratio of each server. Third, to provide necessary traffic isolation for VMs that share the same physical network adapter, we propose the Flow-level Bandwidth Provisioning (FBP) algorithm. By reducing the flow scheduling problem to multiple stages of packet queuing problems, FBP guarantees the provisioned bandwidth and delay performance for each flow. Finally, while DCNs are typically provisioned with full bisection bandwidth, DCN traffic demonstrates fluctuating patterns, we propose a joint host-network optimization scheme to enhance the energy efficiency of DCNs during off-peak traffic hours. The proposed scheme utilizes a unified representation method that converts the VM placement problem to a routing problem and employs depth-first and best-fit search to find efficient paths for flows.

Ecosystem structure in disturbed and restored subtropical seagrass meadows

Shallow seagrass ecosystems frequently experience physical disturbance from vessel groundings. Specific restoration methods that modify physical, chemical, and biological aspects of disturbances are used to accelerate recovery. This study evaluated loss and recovery of ecosystem structure in disturbed seagrass meadows through plant and soil properties used as proxies for primary and secondary production, habitat quality, benthic metabolism, remineralization, and nutrient storage and exchange. The efficacy of common seagrass restoration techniques in accelerating recovery was also assessed. Beyond removal of macrophyte biomass, disturbance to seagrass sediments resulted in loss of organic matter and stored nutrients, and altered microbial and infaunal communities. Evidence of the effectiveness of restoration actions was variable. Fill placement prevented additional erosion, but the resulting sediment matrix had different physical properties, low organic matter content and nutrient pools, reduced benthic metabolism, and less primary and secondary production relative to the undisturbed ecosystem. Fertilization was effective in increasing nitrogen and phosphorus availability in the sediments, but concurrent enhancement of seagrass production was not detected. Seagrass herbivores removed substantial seagrass biomass via direct grazing, suggesting that leaf loss to seagrass herbivores is a spatially variable but critically important determinant of seagrass transplanting success. Convergence of plant and sediment response variables with levels in undisturbed seagrass meadows was not detected via natural recovery of disturbed sites, or through filling and fertilizing restoration sites. However, several indicators of ecosystem development related to primary production and nutrient accumulation suggest that early stages of ecosystem development have begun at these sites. This research suggests that vessel grounding disturbances in seagrass ecosystems create more complex and persistent resource losses than previously understood by resource managers. While the mechanics of implementing common seagrass restoration actions have been successfully developed by the restoration community, expectations of consistent or rapid recovery trajectories following restoration remain elusive.

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.