The Experts below are selected from a list of 519 Experts worldwide ranked by ideXlab platform

Sanger Richard - One of the best experts on this subject based on the ideXlab platform.

  • An algorithmic approach to Openflow ruleset transformation
    The University of Waikato, 2020
    Co-Authors: Sanger Richard
    Abstract:

    In an ideal development cycle for an Openflow application, a developer designs a pipeline to suit their application's needs and installs rules to that pipeline. Their application will run on any Openflow switch, whether software or hardware based. A network operator deploying this application would assess their network's requirements and purchase Openflow hardware to meet these requirements; such as bandwidth, port density, and flow table size. In reality, this level of interoperability does not exist as many Openflow switches are built on a fixed-function pipeline. Fixed-function pipelines limit the matches and actions available to rules depending on the table, but in doing so make more efficient use of expensive hardware resources such as TCAM. This thesis investigates improving Openflow Device interoperability by developing a method to rewrite existing rulesets to new complex fixed-function pipelines. Additionally, this thesis developed the tools to assess and verify the interoperability and equivalence of Openflow rulesets and pipelines. This thesis developed a library and tools for working with descriptions of fixed-function pipelines, specifically, the Table Type Pattern description. This library provides a method to check if an existing ruleset is compatible with a new pipeline. Additionally, this thesis designed and implemented a pragmatic approach to compare if the forwarding behaviour of two Openflow 1.3 rulesets is equivalent. Equivalence checking provides a tool to verify that an Openflow application rewritten to program a new pipeline maintains the correct forwarding behaviour. Finally, this thesis investigates the problem of algorithmically rewriting an existing Openflow ruleset, programmed by an existing application, to fit a different fixed-function pipeline. Solving this problem allows an Openflow application to be written once and run on any Openflow switch. This research aimed to solve this problem in a comprehensive manner that did not rely on the target pipeline supporting features such as Openflow metadata. This thesis developed and implemented a general method to convert an Openflow 1.3 to a complex constrained fixed-function

Japinder Singh - One of the best experts on this subject based on the ideXlab platform.

  • implementation of server load balancing in software defined networking
    2016
    Co-Authors: Sukhveer Kaur, Japinder Singh
    Abstract:

    Network management is very painful and tedious process in large networks having hundreds of switches and routers. Software Defined Networking (SDN) is a new way for creating, designing and managing networks which aims to change this current undesirable situation. The main idea of SDN consists in logically centralizing network control in a SDN controller, which controls and monitors the behavior of the network. In this paper we developed an SDN application that performs server load balancing. The Main problem with traditional load balancer is that they use dedicated hardware. That hardware is very expensive and inflexible. Network Administrators cannot write their own algorithms since traditional load balancers are vendor locked. Therefore to solve these problems, we created SDN application which turned simple Openflow Device into powerful load balancer. There are already certain existing load balancing algorithms in SDN but main problem with all these algorithms is that every request and return message has to pass through the load balancer. It introduces unnecessary latency. To solve these problems we implemented direct routing based load balancing algorithms. In Direct Routing, the load balancer is not involved in the return message from the web server to the client. It means server responds directly to client bypassing the load balancer thus improving performance.

Sukhveer Kaur - One of the best experts on this subject based on the ideXlab platform.

  • implementation of server load balancing in software defined networking
    2016
    Co-Authors: Sukhveer Kaur, Japinder Singh
    Abstract:

    Network management is very painful and tedious process in large networks having hundreds of switches and routers. Software Defined Networking (SDN) is a new way for creating, designing and managing networks which aims to change this current undesirable situation. The main idea of SDN consists in logically centralizing network control in a SDN controller, which controls and monitors the behavior of the network. In this paper we developed an SDN application that performs server load balancing. The Main problem with traditional load balancer is that they use dedicated hardware. That hardware is very expensive and inflexible. Network Administrators cannot write their own algorithms since traditional load balancers are vendor locked. Therefore to solve these problems, we created SDN application which turned simple Openflow Device into powerful load balancer. There are already certain existing load balancing algorithms in SDN but main problem with all these algorithms is that every request and return message has to pass through the load balancer. It introduces unnecessary latency. To solve these problems we implemented direct routing based load balancing algorithms. In Direct Routing, the load balancer is not involved in the return message from the web server to the client. It means server responds directly to client bypassing the load balancer thus improving performance.

Rawat, Danda B. - One of the best experts on this subject based on the ideXlab platform.

  • Design of High Speed Computer Networks Aimed at Reducing Energy Consumption and Carbon Emission
    Digital Commons@Georgia Southern, 2015
    Co-Authors: Howard, Christianna M., Rawat, Danda B.
    Abstract:

    With the exponential growth of Internet use, many organizations (including Georgia Southern University) use vast amounts of energy/power to operate and cool their network infrastructures and thus produce significant amount of carbon waste. The main focus of this research is to design Openflow based networks that will result in high-speed computer networks and dramatic reductions of energy consumption and carbon emission, without reducing the quality of the service. A Software Defined Network saves energy by turning off links or switches within a network when traffic is low, during non-peak hours. In order to determine how many switches or links may be temporarily turned off, an equation must be defined to calculate the power consumption that remains constant within a network Device and that which is affected by traffic. Based on that equation, we can determine the most influential factors of energy consumption within a network. Then a network is designed to keep the energy consumption ratio (ECR) constant as traffic fluctuates. The ECR is the ratio of power (P) used over time (T) compared to number of bits (M) being transferred, or ECR = P*T/M. The power consumed is calculated by combing different elements of power consumption from the Openflow Device. These include the base power to turn on the networking Device, the power dependent on the configuration of switches, the power used by the controls, and the power dependent on the rules set by Openflow software (Pof). By inserting these elements into the ECR, we can evaluate the elements that would have the greatest effect on the ratio. When the number of bits changes, the most affected element of power consumption is the Pof. Therefore the ECR that needs to remain constant will be the ECR_OF = Pof*T/M. An experimental Openflow based network will be established and tested in the CWiNs lab of Georgia Southern University with the specific purpose of demonstrating the significant energy savings and reduction in carbon waste that can be achieved with this ratio while providing reliable and fast networking services at GSU. This research is supported in part by the Center for Sustainability at Georgia Southern University

Howard, Christianna M. - One of the best experts on this subject based on the ideXlab platform.

  • Design of High Speed Computer Networks Aimed at Reducing Energy Consumption and Carbon Emission
    Digital Commons@Georgia Southern, 2015
    Co-Authors: Howard, Christianna M., Rawat, Danda B.
    Abstract:

    With the exponential growth of Internet use, many organizations (including Georgia Southern University) use vast amounts of energy/power to operate and cool their network infrastructures and thus produce significant amount of carbon waste. The main focus of this research is to design Openflow based networks that will result in high-speed computer networks and dramatic reductions of energy consumption and carbon emission, without reducing the quality of the service. A Software Defined Network saves energy by turning off links or switches within a network when traffic is low, during non-peak hours. In order to determine how many switches or links may be temporarily turned off, an equation must be defined to calculate the power consumption that remains constant within a network Device and that which is affected by traffic. Based on that equation, we can determine the most influential factors of energy consumption within a network. Then a network is designed to keep the energy consumption ratio (ECR) constant as traffic fluctuates. The ECR is the ratio of power (P) used over time (T) compared to number of bits (M) being transferred, or ECR = P*T/M. The power consumed is calculated by combing different elements of power consumption from the Openflow Device. These include the base power to turn on the networking Device, the power dependent on the configuration of switches, the power used by the controls, and the power dependent on the rules set by Openflow software (Pof). By inserting these elements into the ECR, we can evaluate the elements that would have the greatest effect on the ratio. When the number of bits changes, the most affected element of power consumption is the Pof. Therefore the ECR that needs to remain constant will be the ECR_OF = Pof*T/M. An experimental Openflow based network will be established and tested in the CWiNs lab of Georgia Southern University with the specific purpose of demonstrating the significant energy savings and reduction in carbon waste that can be achieved with this ratio while providing reliable and fast networking services at GSU. This research is supported in part by the Center for Sustainability at Georgia Southern University