Middleware 05  
3rd International Workshop on Middleware for Grid Computing

Co-located with Middleware 2005, Grenoble, France
Monday, November 28-29th 2005

MONDAY - NOVEMBER 28
Time Title
Slides Text
9:30AM Dependability and Service Discovery
 
R Camargo, R Cerqueira, F Kon
   slides
 
A Banerjee, S Basu, S Garg, S Lee, P Mullan, P Sharma

 slides  
10:30AM  Coffee-Break
11:00AM  Grid Application Frameworks
 
L d''Orazio, F Jouanot, C Labbé, C Roncancio

 slides  
 
M Desertot, C Escoffier, D Donsez

 slides  
12:00  Lunch
14:00  Programming Models and Tools in Grid Middleware
 
A Goldchleger, A Goldman, U Hayashida, F Kon

 slides  
 
A Paventhan, K Takeda

 slides  
 
N Rodriguez, A Milanes, B Schulze

 slides  

 Agent-based Approaches

L Joita, O Rana, O Ardaiz, P Chacin, I Chao, F Freitag, L Navarro

slides
16:00  Coffee-Break
16:30 Peer to Peer Protocols in Grid Middleware - Chair: TBD
 
M Amoretti, F Zanichelli, G Conte

 slides  
 
R Santos, A Andrade, W Cirne, F Brasileiro, N Andrade

 slides  
 
L Lima, A T A Gomes, A Ziviani, M Endler, L F Soares, B Schulze

 slides  
18:00  Summarisation Session
TUESDAY- NOVEMBER 29
Time Title
Slides Text
9:30 Resource Management and Scheduling

V Yarmolenko, R Sakellariou

slides


slides
10:30  Coffee-Break and Posters
11:00
P Vargas, I Dutra, V D do Nascimento, L A S Santos, L da Silva, C Geyer, B Schulze

slides

Portals and Security

A Akram, D Chohan, R Allan

slides

X Zhao, K Borders, A Prakash

slides
12:30  Summarisation Session
12:40  End of Workshop
   Posters
 
J Pathak
   
 
L F Bittencourt , E R M Madeira, F Cicerre, L E Buzato
 slides  
 
A Yamin, I Augustin, C Geyer, L da Silva, R Real
   
 
M J Lima, A L Moura, C Ururahy, T Melcop, M Nery, B Silvestre, V Reis, R Cerqueira
   

Strategies for Storage of Checkpointing Data Using Non-dedicated Repositories on Grid Systems
R Camargo, R Cerqueira, F Kon
  Abstract: Dealing with the large amounts of data generated by long-running parallel applications is one of the most challenging aspects of Grid Computing. Periodic checkpoints might be taken to guarantee application progression, producing even more data. The classical approach is to employ high-throughput checkpoint servers connected to the computational nodes by high speed networks. In the case of Opportunistic Grid Computing, we do not want to be forced to rely on such dedicated hardware. Instead, we want to use the shared Grid nodes to store application data in a distributed fashion. In this work, we evaluate several strategies to store checkpoints on distributed non-dedicated repositories. We consider the tradeoff among computational overhead, storage overhead, and degree of fault-tolerance of these strategies. We compare the use of replication, parity information, and information dispersal (IDA). We used InteGrade, an object-oriented Grid middleware, to implement the storage strategies and perform evaluation experiments.
 
Scalable Grid Service Discovery Based on UDDI
A Banerjee, S Basu, S Garg, S Lee, P Mullan, P Sharma
  Abstract: Efficient discovery of grid services is essential for the success of grid computing. The standardization of grids based on web services has resulted in the need for scalable web service discovery mechanisms to be deployed in grids Even though UDDI has been the de facto industry standard for web-services discovery, imposed requirements of tight-replication among registries and lack of autonomous control has severely hindered its widespread deployment and usage. With the advent of grid computing the scalability issue of UDDI will become a roadblock that will prevent its deployment in grids. In this paper we present our distributed web-service discovery architecture, called DUDE (Distributed UDDI Deployment Engine). DUDE leverages DHT (Distributed Hash Tables) as a rendezvous mechanism between multiple UDDI registries. DUDE enables consumers to query multiple registries, still at the same time allowing organizations to have autonomous control over their registries. We have built and tested a prototype of DUDE on Planetlab. Based on preliminary studies, we believe that DUDE architecture can support effective distribution of UDDI registries thereby making UDDI more robust and also addressing its scaling issues. Furthermore, The DUDE architecture for scalable distribution can be applied beyond UDDI to any Grid Service Discovery mechanism.
 
Building Adaptable Cache Services
L d''Orazio, F Jouanot, C Labbé, C Roncancio
  Abstract:  Caching is crucial to improve performances in many computing systems. It is context dependent, thus many types of cache exist. As a consequence, when a cache is required, it is usually built from scratch. Such a solution is time (and money) consuming, in particular in grid context where several caches may be required. This paper proposes ACS, a framework which allows building adaptable cache services. We present a generic cache definition and provide a description of models implemented in the framework.
 
Autonomic Management of J2EE Edge Servers
M Desertot, C Escoffier, D Donsez
  Abstract: Nowadays, one of the biggest challenges for companies is to cope with the high cost of their information technologies infrastructure. Edge computing is a new computing paradigm designed to allocate on-demand computing and storage resources. Those resources are web cache servers scattered over the ISP backbones. We argue that this paradigm could be applied for on-demand full application hosting, helping reduce cost. In this paper, we present a J2EE dynamic server able to deploy/host J2EE applications on demand and its autonomic manager. For that, we reengineer and experiment JOnAS, an open-source J2EE static server. Two management policies of the autonomic manager were stressed by a simulation of a worldwide ISP network.
 
The Implementation of the BSP Parallel Computing Model on the InteGrade Grid Middleware
A Goldchleger, A Goldman, U Hayashida, F Kon
  Abstract: InteGrade is an object-oriented grid middleware infrastructure whose goal is to leverage existing computational resources in organizations. Rather than relying on dedicated hardware such as reserved clusters, InteGrade focuses on using desktops in users' offices, machines in computer laboratories, shared workstations, as well as dedicated clusters. In this paper, we describe the support for the execution of highly coupled parallel applications on top of InteGrade. The paper describes the implementation of the middleware to support BSP parallel applications (with global synchronization points), and presents experimental results.
 
MyCoG.NET: Towards a mutli-language CoG Toolkit
A Paventhan, K Takeda
  Abstract: Grid applications use Commodity Grid Toolkit (CoG) to consume Globus Grid services. Existing CoG Toolkits are language-specific and are available for Java, Python, and the Matlab scripting environment. In this paper we describe a multi-language CoG toolkit, MyCoG.NET, based around the Microsoft .NET Framework. MyCoG provides set of classes and APIs for .NET. We demonstrate its programmability using FORTRAN, C++, C# and Java, and discuss its performance over LAN and WAN infrastructures. We also demonstrate its applicability by providing a case-study in the field of experimental aerodynamics. We believe this is the first commodity grid toolkit for Windows Platform supporting multi-language. Considering the numerous desktops/laptops running Windows, MyCoG offers users with a platform of choice and extend Grid services reach. Further, it obviates the need for new CoG toolkits to be developed for languages already supported under the .NET framework.
 
Managing Jobs with an Interpreted Language for Dynamic Adaptation
N Rodriguez, A Milanes, B Schulze
  Abstract: In this paper we explore the advantages of using interpreted languages for building submission engines for the grid. In particular, we discuss an example engine, developed using ALua, for the submission of jobs in a cluster, which can be extended to a grid environment. We claim that the flexibility offered by interpreted languages justifies the problems related to the intrinsic loss of efficiency associated to this kind of languages. The focus of this work is on adaptation and ease of use.
 
Application Deployment using Catallactic Grid Middleware
L Joita, O Rana, O Ardaiz, P Chacin, I Chao, F Freitag, L Navarro
  Abstract: In this paper we describe an application deployment using a Catallactic Grid-enabled middleware, which is based on the concept described by von Hayek. The implementation makes use of Globus Toolkit, JXTA and WSRF. The paper envisages the resource virtualization in the WSRF context as the main driver for a proper connection middleware-base platform (on the broad scenario of grid applications).
 
SP2A: a Service-oriented Framework for P2P-based Grids
M Amoretti, F Zanichelli, G Conte
  Abstract: Service-Oriented Architectures (SOAs) are rapidly becoming the key approach for achieving new levels of interoperability and scalability in the development of Grid applications. Within SOA approaches, current solutions for advertising service providers and for allowing prospective clients to discover them are mostly based on centralized registries. Envisioning Virtual Organizations in which all participants are both resource providers and consumers, in a peer-to-peer fashion, seems to be an intereseting solution. In this paper we propose the Service-oriented Peer-to-Peer Architecture (SP2A), a framework enabling peer-to-peer resource sharing in Grid environments. The internal modules of SP2A-based peers are designed as a services, i.e. each one provides a well-defined, self-contained functionality, and does not depend on the context or state of other services. Moreover, the designed infrastructure allows peers to share their resources by publishing the related resource provision service interfaces. The framework has been implemented as a Java API, which currently supports a number of important technologies such as JXTA (peer-to-peer routing), Web Services (service deployment), and OWL-S (semantic description of services).
 
Accurate Autonomous Accounting in Peer-to-Peer Grids
R Santos, A Andrade, W Cirne, F Brasileiro, N Andrade
  Abstract: We here present and evaluate an autonomous accounting scheme that provides accurate results even when the parties (consumer and provider) do not trust each other. Our accounting scheme relies on the observed relative performance among the parties. It is totally autonomous in the sense that it uses only local information, i.e. there is no exchange of information between the parties. This allows for the deployment of the autonomous ac-counting without requiring any sort of identification infrastruc-ture, such as certificate authorities. The no need of trust or so-phisticated infrastructure make our accounting scheme a perfect fit for peer-to-peer grids, which aim to scale much further than traditional grids by allowing free unidentified entry into the grid. Our results show that the proposed scheme performs very close to a perfect accounting scheme whose implementation is unfeasible in most systems, including those we target. Although our autono-mous accounting scheme was developed to work with OurGrid (a peer-to-peer grid that currently comprises around 450 machines spread through about 25 sites, and is in production since Decem-ber 2004), it can also be useful for other systems. The basic re-quirement to use our accounting scheme is that resource consum-ers must also be resource providers.
 
Peer-to-Peer Resource Discovery in Mobile Grids
L Lima, A T A Gomes, A Ziviani, M Endler, L F Soares, B Schulze
  Abstract: Grids are likely to be the mainstay of distributed computing. Grid was firstly applied as a concept for sharing computing resources among wired nodes, but there has been a growing interest in extending this concept to mobile environments. Although significant work has been done towards mobile grids, much of it has followed centralized approaches based on infrastructured wireless networks. We believe that a less restrictive, decentralized approach that supports mobile collaboration in ad hoc wireless networks can cater for novel grid applications. To address this issue, we propose a middleware architecture called MoGrid. MoGrid orchestrates the distribution of grid tasks among mobile devices in a peer-to-peer (P2P) fashion. In this paper, we focus on the P2P Discovery Protocol (P2PDP), which is a central element of our architecture. P2PDP aids in distributing tasks among the most resourceful devices, while mitigating the overhead of control messages exchanged among them. We describe a prototype implementation of our architecture and discuss some issues related to the adoption of P2PDP as an ad hoc resource discovery mechanism in mobile grids.
On the Flexibility of WS-Agreement for Job Submission
V Yarmolenko, R Sakellariou
  Abstract: This paper considers extensions to the WS-Agreement specification, namely the Guarantee Terms of WS-Agreement. Experiences and conclusions drawn are in the context of Agreement-based job management systems. The general idea of these extensions is the use of functions for Guarantee Terms of the Agreement rather than constant values or ranges. Functions may contain variables defined in a particular agreement or be drawn from the known set of reference variables, such as wall-clock time, job start time, etc. We show that such an approach can potentially reduce the negotiation overheads associated with job renegotiation and/or reduce the number of failed agreements.
 
Hierarchical Self-Scheduling for MPI Applications Executing in Computational Grids
C Boeres, A Nascimento, V Rebello, A Sena
  Abstract: The execution of distributed applications on the grid is already a reality. As both applications and grid scale, efficient execution on the available but shared heterogeneous grid resources will be essential. The EasyGrid middleware is a hierarchically distributed Application Management System (AMS) embedded into MPI applications to facilitate their efficient execution in computational grids. The overhead of employing an AMS for each application does however bring at least two benefits. First, the (scheduling) policies adopted can be tailored to the specific needs of its application. Second, distributing the management effort amongst the applications themselves makes grid computing more scalable. This paper describes the low intrusion architecture and implementation of the hybrid scheduling strategy utilized to cope with dynamic grid environments. Using application-specific scheduling policies, near-optimal runtimes highlight the advantages of self-scheduling when executing one or more AMS applications on a grid.
 
Hierarchical submission in a grid environment
P Vargas, I Dutra, V D do Nascimento, L A S Santos, L da Silva, C Geyer, B Schulze
  Abstract: One of the challenges in grid computing research is to provide means to automatically submit, manage, and monitor applications which spread a large number of tasks. The usual way of managing these tasks is to represent each one as an explicit node in a graph, and this is the approach taken by many grid application monitoring systems up to date. This approach can quickly saturate the machine where the application is launched, as we increase the number of tasks. In this work we present and validate a novel architectural model, GRAND (Grid Robust ApplicatioN Deployment), whose main objective is to deal with the problem of memory and load saturation of the submission machine. GRAND is implemented at a middleware level, aiming at providing a distributed task submission through a hierarchical organization. This paper provides an overview of the GRAND submission model as well our implementation. Initial results show that our approach can be much more effective than other approaches in the literature.
Grid Middleware Portal Infrastructure
A Akram, D Chohan, R Allan

Abstract: This paper gives a description of building a Grid middleware portal infrastructure in CCLRC to allow computational scientists, researchers and application users access to resources via an easy to use Web based portal interfaces. The goal is to develop common Grid application components that can be used by portal developers and administrators to build and deploy on open source portal frameworks. This will allow users to authenticate securely to remote resources and provide transparent access via Grid related tools to manage their tasks efficiently. The current portal technology, JSR 168, is described for the development of Java based applications. The work is substantiated by giving an example of building the UK National Grid Service (NGS) portal. NGS Portal itself works as middleware; it glues different Grid Resources together in a single location. Portals and portlets are not confined to a Web-based solution, through Web Services for Remote Portlets WSRP [8] portlets offered by different portal framworks can be consumed in software ranging from desktop to PDA.
SVGrid: A Secure Virtual Environment for Untrusted Grid Applications
X Zhao, K Borders, A Prakash
  Abstract:  Most grid security researches focus on user authentication and secure communication, the protection of grid computers is left to the underlying operating system. Unfortunately, most operating systems protection mechanisms can be turned off after an attacker manages to exploit a vulnerability to gain privileged access. This paper proposes SVGrid, a Secure Virtual Grid running environment to protect filesystem and network from be maliciously accessed even in a compromised environment, and thus allows allow a grid computer to run untrusted grid application. SVGrid works by isolating grid applications in one or more grid virtual machines. The grid virtual machines filesystem and network service was moved into a dedicated monitor virtual machine. All file and network access requests are then forced to go through the monitor virtual machine, where security policies can be enforced in a non-bypassable manner. Because the monitor virtual machine is isolated from grid virtual machine, it cannot be disabled even if a grid virtual machine is compromised. We tested SVGrid against attacks on grid virtual machine using rootkit and internet worm, SVGrid was able to prevent both of them from maliciously accessing filesystem and network. We also evaluated the performance of SVGrid system and found that performance cost was reasonable considering the security benefits of SVGrid.
 

Dynamic Resource Management on the Grid
J Pathak
  Abstract: A grid computing environment enables sharing of loosely coupled resources and services required by various applications in a large-scale. In such an environment, one of the key challenges is to develop a flexible, scalable, and self-adaptive resource management system which would allow users to carry out their jobs by transparently accessing autonomous, distributed, and heterogeneous resources. In this paper, we discuss the research issues and conceptual architectural design of such a dynamic resource management framework, which leverages the open-source Globus Toolkit and commercially available HP OpenView Configuration Management Solutions software (Radia). Our approach provides adaptive and scalable middleware for static and dynamic resource provisioning, resource monitoring, virtual organization-wide authorization, and business policy management. The framework is based on automated, policy-driven change and configuration management functionality that can dynamically adjust the size, configuration and allocation of various resources that will be consumed in the environment.
A Path Clustering Heuristic for Scheduling Task Graphs onto a Grid
L F Bittencourt, E R M Madeira, F Cicerre, L E Buzato
  Abstract: Task scheduling is an NP-Complete problem, in general. Efficient scheduling is very important for achieving good performance for applications executing in heterogeneous computing environments. In this paper we propose a new heuristic for scheduling task graphs onto a grid computing infrastructure that supports dependent task execution. The baseline of the proposed algorithm, called Path Clustering Heuristic (PCH) algorithm, is to select a path from the DAG and schedule the nodes on this path onto the same processor, aimming to reduce communication.
Middleware to Support Pervasive Grid Applications
A Yamin, I Augustin, C Geyer, L da Silva, R Real
  Abstract: The ISAM project primarily seeks for alternatives to build large-scale pervasive applications. It is composed by two subproject: ISAMadapt, that provides tools and abstractions to design and to implement applications; and EXEHDA, that provides a middleware to manage the environment's resources and applications. We consider that the designed middleware is also suitable to build the new generation of grid applications that will assist to mobile users in accomplishing their activities: namely, the pervasive grid applications. This paper presents a general view of the middleware's features based on the description of a pilot application. The main concept embedded in the middleware and application design is the context-awareness expressed by means of built-in adaptive behaviors. This concept is also the key to deal with constraints and unpredictability of the large-scale pervasive environment.
CSBase: A Framework for Building Customized Grid Environments
M J Lima, A L Moura, C Ururahy, T Melcop, M Nery, B Silvestre, V Reis, R Cerqueira
  Abstract: This paper describes CSBase, a framework that allows the development of customized grid environments. CSBase provides end-users with a simplified access to the grid resources and applications. Although all CSBase instances share some common services, such as user and algorithm management services, each instance has its own web interface and specialized applications and services. CSBase is based on the experience gathered from projects developed by our research group in collaboration with industrial partners. This paper also describes some of these projects.