GSoC/GCI Archive
Google Summer of Code 2011

The Globus Alliance

Web Page: http://dev.globus.org/wiki/Google_Summer_of_Code_2011_Ideas

Mailing List: http://dev.globus.org/wiki/Mailing_Lists

The Globus Alliance is a community of organizations and individuals developing fundamental technologies behind the "Grid," which lets people share computing power, databases, instruments, and other on-line tools securely across corporate, institutional, and geographic boundaries without sacrificing local autonomy. Since its creation in 1996, the Globus Alliance has been committed to developing open source software, particularly the Globus Toolkit (http://www.globus.org/toolkit/). In the last year, The Globus Alliance has also launched a new project, Globus Online, a cloud-hosted online service that allows organization to obtain grid capabilities without complex software installation, configuration, and administrative hassles. Globus Online currently provides a transfer service that can be used to move big data securely and reliably among local and distributed computing resources, with more services to follow in the coming year.

Our code repository can be found here: http://code.google.com/p/google-summer-of-code-2011-globus/

Projects

  • Adaptive Tuning in Globus Online The project aims at design of an automatic tuning framework for GridFTP, the primary protocol used within Globus Online for performing bulk data transfers. It involves collecting and analysing data about how round trip time and packet loss rate varies with number of threads used for a TCP connection. Basing on the data, a decision framework shall be implemented. The decision framework will be used to determine optimum thread level in GridFTP.
  • Android App for Globus Online The project is to develop an Android application for Globus Online serivce.
  • Associative arrays and Map-Reduce in Swift Swiftscript is a parallel scripting language that allows large datasets to be processed over a distributed system easily. This project focusses on two areas, Associative arrays and Map-Reduce. Swiftscript currently supports simple arrays accessed by an index. This project extends Swiftscript by introducing associative arrays to the language. We extend the language to support the use of types such as string, boolean and compound types as subscripts. Using the associative arrays we add language support for Map-Reduce style functions.
  • Dead VM Reaper A fault tolerance detector functionality that periodicaly query backend nodes for VM status. This mechanis will update those VM status at the Nimbus frontend's database and avoid the problem of VM that unexpectedly died but is still considered as running by the frontend.
  • Deployment of DemoGrid on FutureGrid resources This project aims to investigate the feasability of providing DemoGrid and its encompassing Globus services as part of the FutureGrid project and facilitate its deployment and use on FutureGrid IaaS clouds for the broader scientific community.
  • Improved Admin Tools This proposal is to provide command line tools for administrator support on the Nimbus clouds. Some examples are terminating a running VM and other things. This will be implemented with a clean API and good documentation.
  • LanTorrent Compression The projects is about enhancing the LANTorrent behaviour while propagating virtual machine images from the central repository across a LAN to many virtual machine monitor nodes.
  • Semantics-Oriented Behavior-Empowered Scientific Service Search Engine Recently emerged services computing technology enables scientists to expose data and computational resources as publicly accessible Web services. Scientists may then utilize published Web services to facilitate in building new scientific workflows, which precisely describe multistep procedures to streamline a composition of tasks and the dataflow among them. However, earlier study shows that services are currently poorly reused in an ad hoc style. Aiming to help domain scientists dynamically fin
  • “GRAPH BASED CASSANDRA PARTITIONER” PROJECT PROPOSAL I am interested on "Graph based Cassandra Partitioner" project under "Globus Online" topic, since it fits best to my interests. Currently, I am working at High Performance Computing Lab at BMI Department as a research assistant (bmi.osu.edu/hpc) where I do research on parallel graph algorithms and graph coloring problem. I summarized existing graph partitioning algorithms in the proposal and suggested a reasonable timeline. I have contacted to Tom Howe (mentor) and got a positive feedback.