Posters
The conference features 14 poster presentations.
9:00-17:30 (12:00-13:45 Core Time) , Tuesday, September 30, 2008
10:00-12:00, Wednesday, October 1, 2008
*G-001
Time Features of Computing Components and the Economic Planning of Resource Transactions
Natalia Currle-Linde and Michael Resch
Abstract: The absence of dynamic time parameters in the description of
application software components does not allow the realization of
standard market relations for the Grid. This also inhibits transparency
in the mechanisms of the allocation of Grid resources both for customers
and providers. Nevertheless, a more customer-driven approach of Grid
management is necessary and will have to be worked out. This paper deals
with the introduction into Grid resource allocation of exact times for
the execution of computation tasks and with the resulting changes in the
Grid organization with a view to the efficient execution of complex
applications. With such a strategy it is possible to implement a full
set of traditional market relations and at the same time provide a
flexible, efficient and transparent distribution of Grid resources.
*G-002
Integrating the Common Information Model with MDS4
Iván Díaz Álvarez, Gracia Fernández López, María José Martín,
Patricia González and Juan Tourino
Abstract: The management and monitoring of static and dynamic resources
is a key issue in Grid environments. Information models are an abstract
representation of software and hardware aspects of these resources, a
common and structured representation that allows intra- and inter-Grid
interoperability. Among them, the Common Information Model (CIM),
standardized by the DMTF, is an object-oriented and extensible model.
The aim of this work is to provide a CIM-based query service for
resource management information in Grid systems. This is achieved by
publishing the CIM information on the Monitoring and Discovery System
version 4 (MDS4) component of the Globus Toolkit, and implementing a
query engine that supports queries with arbitrary navigation depth on a
CIM instance tree.
*G-003
Grid Service Hosting on Virtual Clusters
Bobby House, Paul Marshall, Michael Oberg, Henry M. Tufo and Matthew Woitaszek
Abstract: This paper presents an architecture for service hosting on
virtual clusters spanning multiple administrative domains that balances
the requirements of application developers and resource provider system
administrators. The presented architecture and implementation use
virtual machines to simplify the deployment of externally-accessible
persistent Web and Grid services while allowing resource provider system
administrators to monitor hosted virtual machines and perform critical
maintenance when necessary. This approach allows developers full control
of their distributed resources, specifies a mechanism for resource
provider monitoring and intervention, and reduces the barrier for
hosting user-supplied virtual machines on shared resource provider
cyberinfrastructure.
*G-004
A Scalable High-performance Communication Library for Wide-area Environments
Hideo Saito, Ken Hironaka and Kenjiro Taura
Abstract: We report our progress on SSOCK, a scalable high-performance
communication library for wide-area environments. SSOCK has an API
similar to that of the Socket library, but solves the connectivity and
scalability issues involved with WANs. In one experiment, SSOCK was
able to connect 1,262 processes with each other in a 13-cluster
environment with firewalls and NAT, without any of the connectivity and
resource allocation problems that were encountered when the Socket
library was used. In another experiment in which 100 processes
simultaneously tried to establish connections, SSOCK was able to
establish connections between all pairs of processes in 1.2 seconds,
while the Socket library suffered from a large number of packet losses
and timed out after 189 seconds.
*G-005
Model for Dynamic Grain Sizing Through Compound Parallelization for an Optimization
Problem Solving Grid Application
Mohamed Wahib, Asim Munwar, Masaharu Munetomo and Kiyoshi Akama
Abstract: MHGrid (Meta Heuristics Grid), a service oriented grid
application offering global optimization solvers, allows developers to
integrate their solvers and objective functions through an easy-to-use
transparent mechanism . As a consequence of offering such a service,
MHGrid hosts solvers and objective functions implementing diverse
parallelization models leading later to different job grain size. Yet,
the flexibility of the existing grid programming tools is limited if
used individually. This paper proposes a model that uses a combination
of grid parallel programming tools to enable the developers of MHGrid to
specify their own parallelization model. Using this model the end user
can choose the parallel model of the solver/objective function pair
generating the grain size most appropriate to the problem in hand.
*G-006
ADL: An Algorithm Definition Language for SmartGridSolve
Michele Guidolin and Alexey Lastovetsky
Abstract: SmartGridSolve is an extension of GridSolve that expands the
single task map and client-server model of GridRPC by implementing
server to server communication and the mapping of a group of tasks. In
order to accomplish this functionality SmartGridSolve needs a task graph
that highlights tasks' execution order, communication volume and
computation volume for a given group of tasks. This work presents the
Algorithm Description Language (ADL), a language that helps the
application programmer to easily specify a task graph for any given
algorithm. The language is modular, it has a well-defined structure and
its syntax is similar to ``C'' language. This poster paper introduces a
trivial example of SmartGridSolve application and the use of ADL to
build the relative task graph with an overview of the language syntax.
*G-007
GMount: Build Your Grid File System on the Fly
Nan Dun, Kenjiro Taura and Akinori Yonezawa
Abstract: By GMount, non-privilege users can easily build ad-hoc
distributed file systems in seconds on any machines reachable via SSH.
In the wide-area Grid environments, it is scalable with hundreds of
nodes and usable with NAT or firewall. Given the network topology, the
file system can be created in a locality-aware manner such that the
communication is efficient if applications prefer manipulating data
close to it. GMount is built by using SSHFS-MUX and GXP and can be
effortlessly deployed in multiple clusters. Since SSHFS-MUX is FUSE
based file system, existing binaries can seamlessly run on GMount
without modification. In this paper, we present the design and
implementation of GMount, and its evaluation in a large scale Grid
platform with over 300 nodes spread across 11 clusters.
*G-008
IM.Grid, a Grid Computing Approach for Image Mining of High Throughput-High Content
Screening
HongKee Moon and Auguste Genovesio
Abstract: Abstract: Image processing and analysis has become essential
for both cell biology research and drug discovery since the advent of
High Content Screening (HCS) technologies. In this context, the Grid
technology is a good opportunity to solve intensive computing problems
with large data set. In addition, the exploitation of the Grid is not a
simple task for many users because it is difficult to use the Grid in
practical fields. Another important issue is to provide a simple way to
use of Grid resources. In this paper, we present IM.Grid, a grid
computing extension of our in-house image analysis software called IM
(Image Mining) providing capabilities to simultaneously access visual
data located on NAS (Network-Attached Storage) and extract knowledge
from the raw information by customizable image processing pipeline in a
parallel way. A user makes a plug-in designing own image mining pipeline
using specific built-in image processing libraries. Then, the plug-in
becomes an actual processing unit when Grid starts to analyze multiple
images retrieving them from the NAS at a time. The user receives output
results as fast as numbers of computational grids are available. We
apply this method to reduce the image processing and analysis time of
cell biological images for drug discovery within High Throughput-High
Content Screening (HT-HCS) context. Because the processing time grows
dramatically as the image size becomes huge due to many factors like
multi-channel, high resolution and so on. To deal with these
constraints, we propose a high-performance computing environment on .NET
framework that helps to improve productivity not only in developing
phases but also in HT-HCS platforms.
*G-009
Investigation of the DAG Eligible Jobs Maximization Algorithm in a Grid
Tomasz Szepieniec and Marian Bubak
Abstract: A significant influence of heterogeneity and uncertainty of
grid environment on quality of DAG schedules results in a search for new
approaches. One of them is Internet-based computing scheduling approach
and PRIO algorithm for DAG scheduling. In this paper, we present results
of a detailed evaluation of the PRIO algorithms in a heterogeneous
environment in which schedulers may recognize performance of resources.
*G-010
SOAG: Service Oriented Architectured Grids and Adoption of Application specific
QoS Attributes
Mohamed Wahib, Asim Munwar, Masaharu Munetomo and Kiyoshi Akama
Abstract: Since the introduction of grid computing various efforts were
directed towards integrating and combining the grid with SOA
technologies. That was mainly motivated by capitalizing on the widely
desirable properties of SOA technologies. Nevertheless, the combination
of SOA with grid computing in real practice is mainly tangible in the
alignment of grid technologies with Web services. Therefore the effects
were merely noticeable in the application layer and were more likely to
define a SOA model for inner components' interactions. The deficient
combination of grid with SOA needs to be deeply investigated to reach a
comprehensive integration of grid with SOA that starts from the
application layer till lower grid middleware layers. A major impact in
this integration is offering application layer functionalities as
services and thus having application specific QoS attributes originating
from the application layer to be included in job scheduling. This paper
presents integrating Grid computing with SOA into a generic Service
Oriented Architecutered Grid (SOAG) model. A SOAG model for MHGrid —a
service-oriented grid application for solving global optimization
problems— is also briefly introduced.
*G-011
A GEO Grid Implementation for 3D GIS Taiwan
Guey-Shin Chang, Whey-Fone Tsai, FANG-PANG LIN, Charlie Chang and Te-lin Chung
Abstract: A GEO Grid framework based on grid technology, remote sensing
data, and geographic information developed at National Applied Research
Laboratories (NARL), Taiwan is introduced. The presented approach is
initiated by synergy of NARL's core competence on environment monitoring
and disaster reduction techniques which include high-resolution
satellite image processing, virtual reality visualization, grid
computing, and disaster mitigation technology along with the advanced
cyberinfrastructure environment established within NARL. The NARL's GEO
Grid framework is primarily constituted by three layers, i.e.
application module, service interface, and computing/data/sensor grids
which provide a feasible solution for developing the various modules for
different applications. A prototype platform entitled 3D GIS Taiwan
constituted by using 2m resolution FORMOSAT-2 data and 5m resolution
Digital Terrain Model has been modeled and can be displayed in 3D stereo
visualization and in web pages for the island of Taiwan. The demo
application on the disaster reduction management support is conceptually
elaborated as well. The presented approach emphasizing on the synergy of
multidiscipline with cross-field cooperation for geosciences'
application can become a benchmark in echo to implementation of GEOSS.
*G-012
Semantic Resource Monitoring and Discovery with Rule Processing based on
the Time-series Statistical Data
Mirza Said and Isao Kojima
Abstract: This paper presents the new functional extension to a
semantic resource monitoring and discovery system called S-MDS, which we
have developed. This extension enables to record the monitored value
historically and supports several statistical processing functions which
most of semantic web tools could not support. Based on this historical
statistics which is stored into RDF databases, we have developed the
rule processing functions to the S-MDS. By using this new capability,
it is possible to detect various kinds of status of the monitoring
resources, such as to report the anomaly when the current CPU load
exceeded 3 times of the standard deviation from its average of past
2weeks. Flexibility of monitoring events and conditions is achieved
since users can program their own monitoring rules using a general
purpose rule language.
*G-013
Distributed Data Access/Find System with Metadata for Data-Intensive Computing
submission information
Minoru Ikebe, Atsuo Inomata, Kazutoshi Fujikawa and Hideki Sunahara
Abstract: The data-intensive computing generates a huge number of data
in wide area network. The DataGrid technology tries to manage such
distributed data on the Internet to provide the quick and efficient data
search/access mechanism. The difficulties of the data access on DataGrid
systems is caused by the differences in the data management manner and
policy among organizations which manage storage resources. We propose
the new distributed data management system for data-intensive computing.
Especially, we focus on the data attributes. We define the pairs of data
attribute and its values as metadata. In our system, users can be
find/access data with metadata. We have been developing the prototype
systems. We show the usage of our system with the applications.
*G-014
Model-Based Optimization for Data-Intensive Application on Virtual Cluster
Kento Sato, Hitoshi Sato and Satoshi Matsuoka
Abstract: We propose a model-based optimization algorithm that
determines virtual machine(VM) migration strategies, i.e., which VMs
should be migrated to which nodes, while minimizing I/O access costs. We
solves this problem as a shortest path problem of a direct acyclic graph
which minimizes overall data access costs of target file accesses. Our
simulation-based studies suggest that the proposed algorithm can achieve
higher performance than simple techniques, such as ones that never
migrate VMs or always migrate VMs onto the nodes that hold target files.
|