Science Cloud, Thammasat University

science cloud

The Science Cloud project aims to create a Cloud Computing infrastructure to support computational research at Thammasat University. The project is built using the opensource OpenStack Cloud management software. The system is currently being used for scienctific computation, software developments and class room instructions.

Another objective of the Science Cloud project is to create tools to ease OpenStack deployment and optimize its performance. Recently, we have created a set of shell scripts to assist OpenStack Icehouse installation. They are available for download below.

DOWNLOADS:


Virtual Machine Migration and Checkpointing System

vmich

The Virtual machine Migration and Checkpointing (VMiCh) System is a software system that provides resilience and fault tolerance for virtual machines using Thread-based Live Migration (TLM) and Thread-based Live Checkpointing (TLC) mechanisms. In the Figure below, VMiCh’s architecture consists of three basic components:

VMiCh Architecture
VMiCh Architecture

Our preliminary research results can be found in the TLC paper presented at the IEEE NCA 2011 conference.


The Virtual Cluster-based Coordinated checkpointing Protocol (VCCP)

vccp

We have reported our proof of concepts VCCP implementation and experiments in our VCCP paper presented at IEEE Cluster 2009. We are planning to investigate the following new research topics: The integration of TLC mechanism and VCCP protocol The design and development of a novel virtual network that can seamlessly inter-operate with the VCCP protocol and provide scalable data transmission performance among virtual machines.


Previous Projects:

SushiCloud OpenStack Testbed

sushicloud

This project has the following objectives:

This is our blog built and run on SushiCloud.

DOWNLOADS:


The Checkpointing-Enabled Virtual Machine (CEVM)

cevm

The Checkpointing-Enabled Virtual Machine (CEVM) project is a project to create an efficient, hypervisor-based checkpointing mechanism exploiting available resources in multiprocessing environments. We have concluded this project and presented our results in this thesis and the "An Efficient Virtual Machine Checkpointing Mechanism for Hypervisor-based HPC systems" paper.