LNCS 3320 - A Categorized-Registry Model for Grid ... - Springer Link

2 downloads 621 Views 192KB Size Report
Resource Publication and Discovery(GRPD) using software agents. We use Common ... The CRM uses the OGSA grid service model for state- ful resources.
A Categorized-Registry Model for Grid Resource Publication and Discovery Using Software Agents Lei Cao1 , Minglu Li1 , Henry Rong2 , and Joshua Huang2 1

2

Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai 20030, China {clcao, mlli}@sjtu.edu.cn E-Business Technology Institute, The University of Hong Kong, Hong Kong, China {hrong, jhuang}@eti.hku.hk Abstract. This paper presents a categorized-registry model for Grid Resource Publication and Discovery(GRPD) using software agents. We use Common Resource Model(CRM) specification to describe manageable resources as Open Grid Services Architecture(OGSA) services. We also adopt two-level registry mechanism to register corresponding resources of a Virtual Organization(VO) in order to obtain high GRPD efficiency. The ”Index” node of a VO hosts the general registry. Other specific categorized registries are distributed in the VO. A large-scale Grid system may contain many VOs. ”Index” nodes from various VOs connect to each other in Peer-to-Peer(P2P) mode instead of hierarchical mode. Software agent is a powerful high-level tool for modeling a complex software system. It is adopted to implement GRPD efficiently. Keywords: Grid, GRPD, OGSA, VO, Two-level Registry, P2P, Agent.

1

Introduction

Grid technology has emerged to enable large-scale flexible resources sharing among dynamic VOs. Resource discovery in a Grid is a challenging task because of the following Grid features[1, 2]: heterogeneous, dynamic, autonomic and numerous. These characteristics create significant difficulties for traditional centralized and hierarchical resource discovery services. In this paper, we present a categorized-registry model for GRPD using software agents. We use CRM specification to describe manageable resources as OGSA services. A large-scale Grid system may contain many VOs. Each VO has its own ”Index” node that hosts the general registry. Other specific registries are distributed in the VO. All heterogeneous resources in the VO will register themselves with those specific registries in soft-state way via the general registry. This is a two-level registry mechanism by which GRPD functions can 1

This paper is supported by 973 project(No.2002CB312002) of China, ChinaGrid Program of MOE of China, and grand project of the Science and Technology Commission of Shanghai Municipality(No. 03dz15026, No. 03dz15027 and No. 03dz15028).

K.-M. Liew et al. (Eds.): PDCAT 2004, LNCS 3320, pp. 318–321, 2004. c Springer-Verlag Berlin Heidelberg 2004 

A Categorized-Registry Model for GRPD Using Software Agents

319

be efficiently implemented via software agents. “Index” nodes from various VOs connect to each other in P2P mode instead of hierarchical mode. The rest of this paper is as follows. We discuss related work in Section 2. Section 3 presents our categorized-registry model. We conclude in Section 4 with lessons learned and future research plans.

2

Related Work

Condor-G[3] agent uses the Matchmaker to make brokering decisions about where to execute user jobs. Obviously, the Matchmaker becomes the system bottleneck especially when the system’s scale becomes larger. Index Services of Open Grid Services Infrastructure(OGSI) in Globus[4] may be hierarchically included in a higher-level Index Server. It is not easy to devise scalable Grid resource discovery based on centralized or hierarchical mechanism when a large number of Grid hosts, resources, and users have to be managed[5]. Legion[6, 7] uses a Resource Management Infrastructure(RMI) in which the information collection is very similar to the Information Services in Globus. Objects are used as the main system abstraction throughout. But we can use agents as a high level abstraction. Nimrod-G[8] uses the Globus middleware services to discover dynamic resources and dispatch jobs over computational Grids. Nimrod-G has the same shortcomings as Globus by using Globus Monitoring and Discovery Service(MDS). In the Agent-based hierarchy model[9], the agents are responsible for GRPD. They are deployed hierarchically. In the environment where resources need to be distributed frequently, the overhead will be greatly increased and the nodes that locate in higher levels may become the bottleneck.

3 3.1

Categorized-Registry Model for GRPD Resource Description

Grid resources have some forms of state(runtime,configuration, etc) on which management can be performed. In CRM specification[10], they are known as manageable resources. The CRM uses the OGSA grid service model for stateful resources. The CRM defines a single new port type, BaseManageableResource, which must also be implemented by manageable resource services. It extends the GridService port type. Two important Service Data Elements(SDEs) of BaseManageableResource will be used in our mechanism: serviceGroupType and searchProperty. In our mechanism we extend CRM’s base resource types whose manageability port types have been given in Web Services Definition Language(WSDL) files. 3.2

Model Architecture

In a VO, all manageable resources are represented with various port types according to their types. We adopt multiple registries to accommodate coarse-grained

320

L. Cao et al.

Fig. 1. Model’s architecture and Agent’s structure

classified resource services(See Figure 1(a)). But we also use a general registry to accommodate those registries’ metadata, i.e. their GSHs. This is a two-level registry mechanism. The mechanism strengthens both the efficiency of resource discovery and the scalability of the Grid system. When a client submits a query for specific resources, the query will be submitted to the general registry at first. After being verified with the requested resource type, the query will be delivered to corresponding specific registry. We can benefit much from doing so: (1)decreasing the centralized registry’s workload; (2)strengthening the system’s adaptability; (3)improving the system’s scalability. We use software agents on behalf of these manageable resources to implement GRPD functions. 3.3

Super-Index Network

Use of P2P protocols is expected to improve the efficiency and scalability of large-scale Grid systems. The Grid system may contain many VOs. Every VO has its own Index node to host the general registry we mentioned above. We connect those Index nodes from different VOs into a ”Super-Index network” in P2P manner. Certainly, the Index node will contain the neighbor set of other VOs’ Index nodes. Because the Index node is prone to be a single point of failure and a potential bottleneck of its VO, we take some measures to avoid it: (1)Only host the general registry in the Index node to decrease the workload of the Index node; (2)Use these nodes having the largest capabilities within each VO as the Index node; (3)Introduce redundancy into the design of the Index node to provide more reliability to the VO and less workload on the Index node. However, doing so will add additional cost, so it’s important to balance reliability and cost. In our model we have used two nodes sharing workload of the Index node to ensure good reliability(See Figure 1(a)). The similar measures are taken for those nodes that host specific registries. 3.4

The Structure of Agent

An agent in our model is viewed as the representative of all manageable resource services within the node at a meta-level of resource management. Each agent is able to cooperate with other agents and thereby provides resource publication

A Categorized-Registry Model for GRPD Using Software Agents

321

and discovery for the scheduling of applications that need to utilize grid resources. An agent can be considered as both the service provider and the service requester. Those agents deployed within the VO’s nodes are homogenous and have hierarchy structure. There are three layers(See Figure 1(b)). The communication layer contains a communication module that acts as an interface to the external environment. The central layer is responsible for the core GRPD functions. It has four components: Agent Information Tables(AITs), Matchmaker, Scheduler and Central Manager. The local management layer is used to perform functions such as application management, resource allocation and resource monitoring. In other means, it’s the direct interface to local resources.

4

Conclusion and Future Work

This paper has described a categorized-registry model for GRPD. We have focused on the grouping of manageable resources to build some specific registries in the VO. Doing so increases both the efficiency of resource discovery and the scalability of the Grid system. Using P2P techniques we construct a Super-Index network among VOs of the Grid system. We also utilize software agent technology to complete GRPD efficiently. In the near future, we plan to do a practical performance evaluation to show that our categorized-registry model can be efficiently used in the Grid environment.

References 1. A. Iamnitchi, I.F.: On fully decentralized resource discovery in grid environments. In: Proceedings of International Workshop on Grid Computing. (2001) 2. Y. GONG, e.a.: Vega infrastructure for resource discovery in grids. Computer Science and Technology 18 (2003) 10 3. J. Frey, e.a.: Condor-g:a computation management agent for multi-institutional grids. In: Proceedings of the 10th IEEE International Symposium on High Performance Distributed Computing. (2001) 4. S. Tuecke, e.a.: Open grid services infrastructure (ogsi) version 1.0. In: Specification of GGF by OGSI-WG. (2003) 5. C. Mastroianni, D. Talia, O.V.: P2p protocols for membership management and resource discovery in grids. In: http://biblio.cs.icar.cnr.it. (2004) 6. S.J. CChapin, e.a.: The legion resource management system. In: Proceedings of Job Scheduling Strategies for Parallel Processing. (1999) 7. A. Natrajan, M.A. Humphrey, A.G.: Grid Resource Management In Legion. Kluwer Academic Publishers, Virginia (2004) 8. R. Buyya, D. Abramson, J.G.: Nimrod/g: An architecture for a resource management and scheduling system in a global computational grid. In: Proceedings of the 4th International Conference on High-Performance Computing in the Asia-Pacific Region. (2000) 9. J. Cao, e.a.: Arms: An agent-based resource management system for grid computing. Scientific Programming 10 (2002) 14 10. E. Stokes, N.B.: Common resource model(crm). In: Specification of GGF by CMM-WG. (2003)